Feasibility of encoding vinyl to hi-def audiio?

[FuriousGeorge]

I told my buddy I'd help him convert his Vinyls to a digital format. This is easy enough to do (or so I've read) at high quality, but since I can never leave well enough alone, I was wondering about grabbing hi-def audio from the vinyls.

I know there is much debate as to weather the average human ear can distinguish between 16-bit/44.1KHz and 24-bit/96Khz. I also know how forum discussions can spiral out of control. I don't wish to have this debate, and if you do please try to stop yourself from responding. :rolleyes:

I just want to know if it's feasible (not futile) to actually capture the stream in hi-def, and if so to get some recommendations as to components of this system...

Working from the Source...

The Player:
My buddy spent a lot of money on a Vinyl Player, so I assume it's "good enough". I don't have the Make/Model/Specs on me right now, but if this is a matter of debate I'd be happy to grab them. I know it's on the lower end as far as Direct Drive models go. It has only RCA-Out which brings me to:

The Cables:
I haven't been able to find anything that says that RCA cables are not suited for HD-Audio. I just assume get some 'good cables' and make it happen. Does anyone have any recommendations as to a brand?

The Phono Preamp:
We need one of these. Is a model in the $50-$100 range 'good enough' or should we go higher? Any recommendations?

The Cables (again):
I assume the requirements coming out of the preamp are the same as above.

The Sound Card:
Obviously we'll need a sound card capable of grabbing 24/96 over the Line In. I've always heard good things about the audigy, but is it really necessary to get that expensive? Any recommendations here?

The CODEC:
I gather that WavPack is the best or among the best codecs, from what I've read, so i guess I'll use that as my codec. Can anyone recommend a better codec, Open Source or otherwise?

The Process:
I'm assuming the process involves capturing a WAV at 24/96, then compressing it with my favorite lossless codec. I will probably do this in Linux with WavPack, but if there is any de facto standard software for Windows used in digitizing Vinyls then please let me know.

Much thanks in advance.

[Viracocha]

Everything you've mentioned is extremely low end, so although this doesn't mean the recordings will be 'bad', you don't need to worry so much about ultra high definition as you aren't using high quality components.

Also, if you are just going to be listening to them as mp3s afterwards, it makes even less difference!

The audigy is just a gaming card, it isn't recommended for decent audio recording use as it locks to 48Khz when using asio drivers (i think this is still true, it was a huge issue before).

Forget the audigy. All you need is a solid no nonsense card with decent converters, and you can get this with the M-audio audiophile 24/96 - a very basic but highly recommended card and perfect for this application.

Your preamp - what was he using before? Do you not have a dj mixer kicking around? That will do the exact same job as it is a preamp itself.
Otherwise yes you'll need a dedicated phono preamp, the more you spend the better the quality, that's life. $100 isn't very much, but then again you shouldbe able to find somethign acceptable so long as it is purely a phono preamp with no other bells and whistles.
All high end consumer audio uses rca sockets as the signal is always unbalanced (2 wires - ground and signal) This is no problem so long as the cables are kept short, 2m is fine.

As for the recording format, I think vinyl has a bandwidth of 25KHz, so if you want to capture every last drop, 88.2KHz is what you want, no point doing 96 as you won't be capturing anything you can't with 88.2, plus there are fewer rounding errors going from 88.2 to 44.1, whether this is a big issue is debatable, but still makes sense.
That's it.
Once you've recorded everything, top and tail the recordings and then if I were you, I'd just leave them the hell alone.
If you want to put them on a portable mp3 player then encode at a minimum of 256kbps, but leave the hi-res versions intact on the hard drive for listening at home, otherwise there's no point at all doing all this in the first place.
Use Lame as your encoder, it's the best I've found for mp3.
And by the way, have you any idea how much time it is going to take to do an entire vinyl collection? :D

[oretez]

I don't really know what you mean by 'hi-def' audio. I have a guess but since I have never found audio accompanying high definition video to particularly appealing . . . Well, as I said I can't be entirely sure what you mean. With that caveat

First rule of thumb is that the analog portion of the signal chain is far more important, sensitive to chaotic fluctuations then is the digital portion.

Generally speaking your original signal source is what determines the quality, is the most important limiting variable for the quality of the final artifact. Even with extensive 'clean up' and editing of digital capture, even when you tip into resynthesis you are essentially applying make up, unless you have original (un mixed raw) files you are not even delving into cosmetic surgery. Reason for expressing the obvious is that far more important then 'codec', audiophile cables (which are in any case voodoo), is/are physical condition surrounding initial capture. Cleaning the platters (and there is a whole encyclopedia of obsessive behavior into which one can delve in that topic, brush, moving air, maybe neutral pH water something that dries without adding (much) additional transfer is probably 'good enough'), making sure the arm tracks accurately, that the stylus is appropriate for the task . . .

Which gets you to the pre amp question, and the way you phrase it more or less answers your 'hi-def' inquiry . . . To not spend time cleaning, adjusting the tracking, balance, weight of the arm and stylus, opting for a consumer grade $50 phono preamp precludes any objective value in worrying about digital portion of the project. If the cheap consumer preamp is fine for you (and there is no reason why it can't be . . . I walk around the corner into a room with in excess of $100K gear designed to let me hear music accurately (whether hi-def or not) but when I listen for relaxation I typically use either a 10 yr. old $30 CD player over speakers that cost less then $100 in the car or 160 kbs MP3 streamed wirelessly over a mono speaker in the house (well the speaker enclosure is bass reflex ported with two drivers but the signal is summed to mono prior to reaching the speaker) I can tell the difference, between listening experiences among the three but I'm not going to acoustically treat every room in the house or spend time fiddling with dials when I just want to listen without thinking so the lo-def walking around speaker works fine for me (and I like it better then ear buds, strategy preferred by my wife)), matches your budget for the project then 'hi def' is unlikely to add anything significant. The scope, size of the project can also be a controlling factor for appropriate gear (budgets are always time+money). With 100 albums any 'x', entry level, gear might be perfectly adequate while for 2000 spending more for gear that while not necessarily superior in under the hood components but which facilitate work flow can be beneficial.

Anyway point is that I am not trying to be snobbish about any particular gear . . . But of all the elements of a project like this hi-def (what ever that might be) vs. Lo-def is the least of your worries.

That said, if you are going to do any extensive editing (for me, that edit threshold would be normalizing the tunes . . . But that is a fairly low threshold) then there are reasons (particularly with muscle available in current computers) to capture not only at 24 bit but work with 32 bit floating point . . . With vinyl as source it is harder to make a case for using anything other then 44.1 kHz And opting for 32 bit does not, directly have much to do with psycho acoustics, quite a lot to do with accuracy (not accumulating rounding errors with successive edits) of math involved with any digital editing.

My rough guess is, based solely on info you provide, is that you would find it very difficult to tell difference, in Blind A/B, between 16/44 and 24/96 capture. Generally speaking 16 bit will exceed the dynamic range of the original recordings, 44 will accurately represent coherent frequencies of the source and exceed capabilities of delivery system (amplifier, speaker, room, ears)

Good luck

[oretez]

this is not an area of expertize for me

on the surface certainly have no complaints about this cost effective device: http://www.phonopreamps.com/TC-760LCpp.html

the types of issues I'd want addressed, and which typically can only be addressed in the location where you are doing the transfers has to do with how current supplying power to various devices is going to interact with the very very low level signal leaving tone arm stylus. Wall warts are notorious for ability to introduce unacceptable noise even into order of magnitude greater line level signals

under the hood specs look OK, price seems reasonable but 'interconnects' are always going to be problematic areas of weakness.


This: http://www.samedaymusic.com/product--ARTUSBPHONOPLUSV2 functions as both phono pre and USB audio capture, never used it but ART does market entry level gear that tends to live under hard use. primary concerns here would still be with inteactions between power supplies and signal. (and I'm sure Sweetwater has same or similar at same or similar price)

[FuriousGeorge]

Thanks so much for the fast response!

Everything you've mentioned is extremely low end, so although this doesn't mean the recordings will be 'bad', you don't need to worry so much about ultra high definition as you aren't using high quality components.


The only component that we actually have and cannot be replaced is the Record Player. It's not that it can't be replaced, but it would be expensive to do so. I thought the Direct Drive Record Players were generally of good quality. I can get you the make/model if it clears anything up.

I'm just looking for suggestions as to everything else so that assuming the record player isn't too crappy for high-def output, nothing else is either.


Also, if you are just going to be listening to them as mp3s afterwards, it makes even less difference!


Correct me if I'm wrong, but if I encode to WavPack I can then reconvert, for instance, to flac in .ogg container and play it at the same quality in any player that supports .ogg.

The number 1 goal is to preserve the records at the best possible quality. Once we have that I see no problem in playing them as .WVs right off the computer as a secondary benefit.

Only a knucklehead would record encode at 24/96 just to go to .mp3. Might as well go right to .mp3 at that point. ;)


The audigy is just a gaming card, it isn't recommended for decent audio recording use as it locks to 48Khz when using asio drivers (i think this is still true, it was a huge issue before).

Forget the audigy. All you need is a solid no nonsense card with decent converters, and you can get this with the M-audio audiophile 24/96 - a very basic but highly recommended card and perfect for this application.


Thanks for the heads-up about the audigy, and the recommendation.


Your preamp - what was he using before? Do you not have a dj mixer kicking around? That will do the exact same job as it is a preamp itself.


He's using a Denon Receiver with a preamp, but it has no 'outs' so we'll need a new one.


Otherwise yes you'll need a dedicated phono preamp, the more you spend the better the quality, that's life. $100 isn't very much, but then again you shouldbe able to find somethign acceptable so long as it is purely a phono preamp with no other bells and whistles.


Sounds like you're saying a high-quality phono preamp suitable for hi-def encoding downstream would run about $100.


All high end consumer audio uses rca sockets as the signal is always unbalanced (2 wires - ground and signal) This is no problem so long as the cables are kept short, 2m is fine.


We were thinking of getting high-quality cables and keeping the distance under 1m on either side of the amp. Are you talking about 2m total or 2m per feed? Either way, we'll keep it short.


As for the recording format, I think vinyl has a bandwidth of 25KHz, so if you want to capture every last drop, 88.2KHz is what you want, no point doing 96 as you won't be capturing anything you can't with 88.2, plus there are fewer rounding errors going from 88.2 to 44.1, whether this is a big issue is debatable, but still makes sense.


Nice nugget on Vinyl's bandwidth, though I must admit I don't understand how you derive the optimal sampling rate from the maximum bandwidth of Vinyl. You don't have to explain, but a link would be nice if you can find the time.

Putting the debate aside, would a higher quality soundcard lower the rounding errors, or is that CPU related? Is there anything that can be done about that, just in case one side of the debate is in fact more righteous?


That's it.
Once you've recorded everything, top and tail the recordings


You just went Greek on me. ;) I don't want to trouble you for an explanation, but a good link is welcome if you can find the time.


and then if I were you, I'd just leave them the hell alone.
If you want to put them on a portable mp3 player then encode at a minimum of 256kbps, but leave the hi-res versions intact on the hard drive for listening at home, otherwise there's no point at all doing all this in the first place.


I have a fairly good understanding of how PCM works... so... duh! ;)


Use Lame as your encoder, it's the best I've found for mp3.
And by the way, have you any idea how much time it is going to take to do an entire vinyl collection? :D

I guess that depends on the software, the hardware, and the number of Vinyls ;)

Time we have. Besides, we have the .WAVs to play with while we recode to .WV so it's not a total loss of time until the final product. Plus we get to listen to the Vinyls in stage one, so it could be fun for all right from the beginning.

Thanks again for the responses, and I would appreciate it if you would chime in again any time you feel like it.

[oretez]

Thanks so much for the fast response!
[snip]

Only a knucklehead would record encode at 24/96 just to go to .mp3. Might as well go right to .mp3 at that point. ;)
[snip]

You just went Greek on me. ;) I don't want to trouble you for an explanation, but a good link is welcome if you can find the time.




again not necessarily (knucklehead) . . . again what you do with digital file after capture determines optimal capture parameters . . . if there is no step between capture and mp3 listening then streamlining the process might be beneficial . . .

top & tail . . . different encoders treat beginning and end of audio capture differently (an area where digital is uniquely differnt then analog) typically an encoder will specify number of samples per second (per ms, per whatever) you can get there by either removing or adding samples . . . if encoding added more samples then you want you can typically trim them

additionally not all encoders are created equal with regard to how initial and terminal are handled. Leaping instantaneously from nothing to something, dropping from something to nothing instantaneously . . . if process stops at a point where sine wave is not crossing '0' will introduce an audible 'click' . . . again these, if they occur, can be edited out relatively easily @ 'top & tail'

anyway 'top & tail' is simply reference to beginning and ending, 'topping & tailing' typically refers to some specific repeated action (though the actiom might differ from project to project, community to community . . . )

[AndyH]

This Denon receiver would be very unusual if it did not have a tape loop. The 'tape out' or 'monitor out' is the place to connect to the soundcard inputs. That is exactly what would be recorded to tape if you put a tape deck there and exactly what is being sent to the amplifier part of the receiver. If that sounds like you want it to, the signal coming out at that output will too.

The only possible problem with the above is signal level. Exactly the same consideration will apply to most stand-alone phono preamps and most certainly to all high quality ones.

My receiver out to Audiophile 2496 level is totally acceptable but cartridge outputs and phono preamp amplification is different for every combination. Therefore, you might get too hot a signal that will clip at the soundcard inputs. If so, you need a mixer, or some form of attenuation, in-between the receiver tape out (or phono preamp out) and the soundcard.

There probably isn't any case where the signal level won't be high enough, so it won't need further amplification before the soundcard, there is just the possibility of it being too high. That said as an expression of fact, some people are emotionally dissatisfied and foolishly fiddle around with trying to get the input as near as possible to 0dB to record at a higher level (using a mixer or line level preamp after the phono preamp). It isn't necessary and is quite unlikely to actually improve anything. However, if that is you, you will need said mixer or preamp regardless of your other equipment, as every LP level is also different.

[oretez]

Nice nugget on Vinyl's bandwidth, though I must admit I don't understand how you derive the optimal sampling rate from the maximum bandwidth of Vinyl. You don't have to explain, but a link would be nice if you can find the time.

Relationship between sampling frequency and maximum frequency that can be represented accurately is described by Nyquist-Shannon theorem. Roughly speaking a 'sample' can accurately represent the original if sampling process is 'band limited' and if sampling frequency is more then twice the original. I.E. 44.1 can present 22kHz accurately, 88, 44kHz, etc. Thus the theoretic 25 kHz maximum frequency of LPs would require a greater then 50 kHz sampling rate. (and in fact some of the earliest digital tape used slightly greater then 50 kHz as a sampling frequency).

Without devolving into sampling warz there are reasons why 44 & 48 became industry standards. For practical purposes it is unlikely that most people (and OP's description of budget and gear tends to make me include him in that category) have delivery systems (amp speakers room ears) that would reproduce coherent signals above 22 kHz even if it were biologically possible (if one is older then 25 highly improbable in any case) to hear them

[Viracocha]

The only component that we actually have and cannot be replaced is the Record Player. It's not that it can't be replaced, but it would be expensive to do so. I thought the Direct Drive Record Players were generally of good quality. I can get you the make/model if it clears anything up.

LIke Andy said, the quality ofyour cartridge and the way the arm is set up will have more of an impact, so as long as you get those right you should be good to go.

Correct me if I'm wrong, but if I encode to WavPack I can then reconvert, for instance, to flac in .ogg container and play it at the same quality in any player that supports .ogg.

The number 1 goal is to preserve the records at the best possible quality.
PCM will give you the quality - I don't actually know what wavpack does, but flac and ogg I'm familiar with, so yes, any lossless compression format is cool and the gang for local storage and you can just stash the original wavs on an external drive or blueray disc or whatever.

Thanks for the heads-up about the audigy, and the recommendation.

No probs.
Although now that Andy's mentioned it, the lack of hardware atennuation before the converters 'may' be a problem, if the rms (average volume, not just peak) level of the material you're recording is high, it may cause distortion at the input stage of the soundcard converters, and of course any peaks going over will clip and that's not good.
Dave (ecc83) on this forum is the man to ask about the attenuator, he's a hardcore electronics buff :)

Also as mentioned above, don't be tempted to get your signl going in really hot to the soundcard for the same reasons, you'll get a cleaner signal with less risk of clipping by going in a bit lower. If you record at 24 bit you won't hit the noisefloor for quite some time so don't be scared to go in lower than you may normally do.
As soon as your peaks hit -6dBFS then you are using the highest tier of resolution, no need to aim for just under 0dB as that will inevitably backfire.

He's using a Denon Receiver with a preamp, but it has no 'outs' so we'll need a new one.

Again, sure you're not missing them because they are labeled differently? It would be good of you could come straight out of the amp into the soundcard as that solves your preamp and attenuation problem in one go, but I guess if there's no connections, there's no connections!

Sounds like you're saying a high-quality phono preamp suitable for hi-def encoding downstream would run about $100.

Hmm. I have no experience of stand-alone phono preamps other than finding one for a friend on a super low budget, so i saw how cheap they can be, I just figured that for 100 bucks you couldn't completely shoot yourself in the foot :D
Careful with the use of the word 'encoding' by the way, normally refers to translation of one digital format to another, you are recording, encoding is when you turn the recordings into flac, mp3 or whatever.

We were thinking of getting high-quality cables and keeping the distance under 1m on either side of the amp. Are you talking about 2m total or 2m per feed? Either way, we'll keep it short.

Yeah, short is good. Unbalanced lines are more susceptible to RFI when iusing longer lengths, 2m is fine (as in 2m left, 2m right).

Nice nugget on Vinyl's bandwidth, though I must admit I don't understand how you derive the optimal sampling rate from the maximum bandwidth of Vinyl. You don't have to explain, but a link would be nice if you can find the time.

I actually had a look around after writng that and it seems that a brand spaning new bit of vinyl will get to about 22KHz, a well worn one can drop down to about 18KHz, so in fact you will be fine with 44.1KHz.
(All you do is halve the samplerate to find the maximum frequency that rate is capable of capturing).

Putting the debate aside, would a higher quality soundcard lower the rounding errors, or is that CPU related? Is there anything that can be done about that, just in case one side of the debate is in fact more righteous?

It's neither. It's just the inherent nature of digital. You can get more accurate rounding by doing any processing to the recorded file in a 32 bit application though. That's why digital audio mix engines in sequencers always operate at a minimum of 32 bits.

You just went Greek on me. ;) I don't want to trouble you for an explanation, but a good link is welcome if you can find the time.

Ah! Just means the boring trimming at start and end points of the tracks. Make sure that 'snap to zero crossing' or similar is activated in whatever program you use to do the editsso you don't get any unexpected clicks and pops.
That's another point actually, clicks and pops. 'Audacity' is a freeware audio editor that you can use to record your files into, edit them, and it also has a built-in wave restorer/click remover that may well come in handy.

I have a fairly good understanding of how PCM works... so... duh! ;)

Sorry :p

Plus we get to listen to the Vinyls in stage one, so it could be fun for all right from the beginning.

That is extremely true! A few mates over, some beers, a big bag of weed and a heap of vinyl, sounds like heaven!
Just don't forget to press record ;)

[FuriousGeorge]

Wow you guys are fast. Oretez, here's my followup to you:


I don't really know what you mean by 'hi-def' audio. I have a guess but since I have never found audio accompanying high definition video to particularly appealing . . . Well, as I said I can't be entirely sure what you mean. With that caveat

First rule of thumb is that the analog portion of the signal chain is far more important, sensitive to chaotic fluctuations then is the digital portion.

Generally speaking your original signal source is what determines the quality, is the most important limiting variable for the quality of the final artifact.


I know what you mean about the source and let me elaborate on the assumptions I'm making that led me to post about this:

ASSUMPTION #1: (According to my buddy) The sound that comes out of a decent LP Player (e.g. his) and good quality LPs (e.g. his) is closer to the original source (the musicians in the studio), than a CD (which, after all, is why he bought the player, and buys records. I don't know if I agree with this assumption but I'm willing to run with it).

ASSUMPTION #2: An enthusiast-grade preamp, cables, and sound card can get this source to the destination (a .wav file on my hard disk) with negligible degradation.

CONCLUSION: I can record the original source at a bit/rate > 16/44.1 and objectively capture a higher definition of audio with meaningful sound-data (i.e. like the LP, also closer to the original source in the studio than a CD).

To be clear I have no doubt that subjectively speaking I will not be able to tell the difference, and I doubt he will either, even with proper output components. The #1 priority is to record the LP with the highest level of fidelity.


Even with extensive 'clean up' and editing of digital capture, even when you tip into resynthesis you are essentially applying make up, unless you have original (un mixed raw) files you are not even delving into cosmetic surgery. Reason for expressing the obvious is that far more important then 'codec', audiophile cables (which are in any case voodoo), is/are physical condition surrounding initial capture. Cleaning the platters [snip], making sure the arm tracks accurately, that the stylus is appropriate for the task . . .

Which gets you to the pre amp question, and the way you phrase it more or less answers your 'hi-def' inquiry . . . To not spend time cleaning, adjusting the tracking, balance, weight of the arm and stylus,


I hadn't even considered remastering. Is there any software you would recommend for this?

We did plan on getting an LP cleaner, but nothing better than the $20 garden variety. Most of his LPs are brand new.

As to the mechanical aspects (stylus, arm, etc), let's assume for now that as it stands they produce a superior sound to a CD. Again, that's what he says. I'm not saying I agree.


...opting for a consumer grade $50 phono preamp precludes any objective value in worrying about digital portion of the project. If the cheap consumer preamp is fine for you (SNIP: ed.: Holy parenthetical statement batman!), matches your budget for the project then 'hi def' is unlikely to add anything significant.


And therein lies the purpose of my post. Pretend you agree with ASSUMPTION 1 for the purposes of this discussion, and if I am correct about ASSUMPTION 2, kind of preamp / soundcard / etc. would I need to objectively (not necessarily subjectively) record a stream that would benefit from bit/rate > 16/44.1.

I mean, we have to assume that 16/44.1 is not the optimal bit/rate for audio, right? Why else would DVD-A be 24/96? Isn't 16/44.1 just something that was reasonable for the digital recording capabilities of the 1980s?


The scope, size of the project can also be a controlling factor for appropriate gear (budgets are always time+money). With 100 albums any 'x', entry level, gear might be perfectly adequate while for 2000 spending more for gear that while not necessarily superior in under the hood components but which facilitate work flow can be beneficial.


We can take time out for now. While different aspects of this project amuse us, it falls squarely under the hobby category for both of us.


Anyway point is that I am not trying to be snobbish about any particular gear . . . But of all the elements of a project like this hi-def (what ever that might be) vs. Lo-def is the least of your worries.


Again, what I'm asking is, based on our hypothetical agreement with my ASSUMPTIONS, what would it take to make it a concern? How much would that cost.


That said, if you are going to do any extensive editing (for me, that edit threshold would be normalizing the tunes . . . But that is a fairly low threshold) then there are reasons (particularly with muscle available in current computers) to capture not only at 24 bit but work with 32 bit floating point . . .


...but there are not enthusiast grade sound cards that can capture a stream at 32-bit, or am I wrong about this.


With vinyl as source it is harder to make a case for using anything other then 44.1 kHz And opting for 32 bit does not, directly have much to do with psycho acoustics, quite a lot to do with accuracy (not accumulating rounding errors with successive edits) of math involved with any digital editing.


I get what you're saying about 32-bit capture and rounding errors, but it sounds like you're saying that a sampling rate > 44.1 is in fact futile (ther by answering my post).

Really? Even if you agree with my assumptions, if only for the purposes of this discussion?


My rough guess is, based solely on info you provide, is that you would find it very difficult to tell difference, in Blind A/B, between 16/44 and 24/96 capture.


I have little doubt that neither of us will be able to tell the diference, subjectively speaking, but keep in mind that our goal is objectively higher fidelity, even if it is never heard or distinguished by either of us. Just doing it for the sake of doing it.


Generally speaking 16 bit will exceed the dynamic range of the original recordings, 44 will accurately represent coherent frequencies of the source and exceed capabilities of delivery system (amplifier, speaker, room, ears)


Does taking the delivery system and our ears out of the picture make any difference?

As to your second response: Thanks for the suggestions on the preamp. I like the first one better than the second one, since (correct me if I'm wrong) the second one would eliminate the need for a soundcard all together, as it would digitally deliver the recording (@ 16/48 max) to the computer, and I like the card Viracocha suggested.




Only a knucklehead would record encode at 24/96 just to go to .mp3. Might as well go right to .mp3 at that point.


again not necessarily (knucklehead) . . . again what you do with digital file after capture determines optimal capture parameters . . . if there is no step between capture and mp3 listening then streamlining the process might be beneficial . . .


I suppose I deserved that for poor wording and tacit implications, but I meant that it made no sense to take your wav->.wv (or flac) at 24/96 if the goal was to go to .mp3 and play it on you ipod.

Thanks for all the other info in that third reply. I'll need time to digest that, to do some homework, and get some practical experience before I can really begin to understand that.

[FuriousGeorge]

This Denon receiver would be very unusual if it did not have a tape loop. The 'tape out' or 'monitor out' is the place to connect to the soundcard inputs. That is exactly what would be recorded to tape if you put a tape deck there and exactly what is being sent to the amplifier part of the receiver. If that sounds like you want it to, the signal coming out at that output will too.


I'll need to verify this with my buddy. It does sound like what we want to do.


The only possible problem with the above is signal level. Exactly the same consideration will apply to most stand-alone phono preamps and most certainly to all high quality ones.

My receiver out to Audiophile 2496 level is totally acceptable but cartridge outputs and phono preamp amplification is different for every combination. Therefore, you might get too hot a signal that will clip at the soundcard inputs. If so, you need a mixer, or some form of attenuation, in-between the receiver tape out (or phono preamp out) and the soundcard.

There probably isn't any case where the signal level won't be high enough, so it won't need further amplification before the soundcard, there is just the possibility of it being too high. That said as an expression of fact, some people are emotionally dissatisfied and foolishly fiddle around with trying to get the input as near as possible to 0dB to record at a higher level (using a mixer or line level preamp after the phono preamp). It isn't necessary and is quite unlikely to actually improve anything. However, if that is you, you will need said mixer or preamp regardless of your other equipment, as every LP level is also different.


I think I understand what you are saying (a little). How do I know if the signal is 'too hot'? Should I listen for distortion, or can I 'see it' using good software. I have no practical experience with recording analog audio since I would dub my CDs to tape in the early 90s.

I do sound like the type to be emotionally dissatisfied, but when does 'too hot' become so hot that it goes beyond emotion?

[dcwave]

My two cents - record the record at 44.1/24 bit. This will give you plenty of fidelity and will not give you headaches with SRC if you decide to go to CD.

[FuriousGeorge]

My two cents - record the record at 44.1/24 bit. This will give you plenty of fidelity and will not give you headaches with SRC if you decide to go to CD.

Thanks for the pennies. The room seems to agree with you. Seems like, to my surprise, Vinyl has a sample rate. I had no idea. See big letters below.

LIke Andy said, the quality ofyour cartridge and the way the arm is set up will have more of an impact, so as long as you get those right you should be good to go.


PCM will give you the quality - I don't actually know what wavpack does, but flac and ogg I'm familiar with, so yes, any lossless compression format is cool and the gang for local storage and you can just stash the original wavs on an external drive or blueray disc or whatever.



No probs.
Although now that Andy's mentioned it, the lack of hardware atennuation before the converters 'may' be a problem, if the rms (average volume, not just peak) level of the material you're recording is high, it may cause distortion at the input stage of the soundcard converters, and of course any peaks going over will clip and that's not good.
Dave (ecc83) on this forum is the man to ask about the attenuator, he's a hardcore electronics buff :)

Also as mentioned above, don't be tempted to get your signl going in really hot to the soundcard for the same reasons, you'll get a cleaner signal with less risk of clipping by going in a bit lower. If you record at 24 bit you won't hit the noisefloor for quite some time so don't be scared to go in lower than you may normally do.
As soon as your peaks hit -6dBFS then you are using the highest tier of resolution, no need to aim for just under 0dB as that will inevitably backfire.


Thanks. That does clear things up. Still, I need to do some homework and get some practical experience, as we are reaching the point where your sage advice is out of my league.



Again, sure you're not missing them because they are labeled differently? It would be good of you could come straight out of the amp into the soundcard as that solves your preamp and attenuation problem in one go, but I guess if there's no connections, there's no connections!


Definitely no connection... Just verified.


Careful with the use of the word 'encoding' by the way, normally refers to translation of one digital format to another, you are recording, encoding is when you turn the recordings into flac, mp3 or whatever.


Lesson learned. I guess I got confused because Analog TV capture cards for computers are often called mpeg encoders, but I guess, now that I think about it, the analog stream is already RECORDED as the video equivalent of a waveform (raw video?) by the time it is ENCODED to mpeg ;)


I actually had a look around after writng that and it seems that a brand spaning new bit of vinyl will get to about 22KHz, a well worn one can drop down to about 18KHz, so in fact you will be fine with 44.1KHz.
(All you do is halve the samplerate to find the maximum frequency that rate is capable of capturing).


Vinyl has a Sample Rate!?! I had no idea... I thought it was kinda like how analog photographs don't have a resolution. I didn't think you could talk about that media in those terms.

I guess that throws a wrench in my assumptions for oretez. (well, at least the part about a sample rate > 44.1)

{{
UPDATE: You guys are responding faster than I can learn. I just read oretez's explanation:


Relationship between sampling frequency and maximum frequency that can be represented accurately is described by Nyquist-Shannon theorem. Roughly speaking a 'sample' can accurately represent the original if sampling process is 'band limited' and if sampling frequency is more then twice the original. I.E. 44.1 can present 22kHz accurately, 88, 44kHz, etc. Thus the theoretic 25 kHz maximum frequency of LPs would require a greater then 50 kHz sampling rate. (and in fact some of the earliest digital tape used slightly greater then 50 kHz as a sampling frequency).

Without devolving into sampling warz there are reasons why 44 & 48 became industry standards. For practical purposes it is unlikely that most people (and OP's description of budget and gear tends to make me include him in that category) have delivery systems (amp speakers room ears) that would reproduce coherent signals above 22 kHz even if it were biologically possible (if one is older then 25 highly improbable in any case) to hear them


... though I maintain it isn't about the delivery system. It's about capturing the original LP with the greatest level of fidelity (without crossing the futility threshold) just for it's own sake, and even if it is never played on a capable delivery system for ears sufficiently acute.
}}

So that leaves the bit-level as my best hope for the highest possible fidelity recording. Speaking of which:


You can get more accurate rounding by doing any processing to the recorded file in a 32 bit application though. That's why digital audio mix engines in sequencers always operate at a minimum of 32 bits.


... but there are no enthusiast grade sound cards capable of > 24-bit capture, are there?

I looked and couldn't find any.


A few mates over, some beers, a big bag of weed and a heap of vinyl, sounds like heaven!
Just don't forget to press record ;)


Sounds like my kinda party. Besides the press we'll need some rolling papers too ;)

[sabianq]

A vinyl record has a groove carved into it that mirrors the original sound's waveform. There is limited headroom on a vinyl record, during recording, to much dynamic range will cause the cutter to cut the grove into the adjacent grove. but there is no such thing as a sample rate on a vinyl record.

a vinyl recording can be more accurate than a recording on a compact disk, a digital recording is not capturing the complete sound wave. It is approximating it with a series of steps (the sample rate).

http://electronics.howstuffworks.com/question487.htm

as for recording, you should be archiving your vinyl recordings at the highest native sample rate your interface can sample at.

44.1/24bit or 48/24bit or 96/24bit

Store the archives as an unprocessed and uncompressed, raw, high resolution digital wave format on DVD.

then use the archive to (down sample) to MP3 or other format.

it is always best to process from a high resolution digital wave to a compressed truncated format like MP3.

cheers!

[FuriousGeorge]

A vinyl record has a groove carved into it that mirrors the original sound's waveform. There is limited headroom on a vinyl record, during recording, to much dynamic range will cause the cutter to cut the grove into the adjacent grove. but there is no such thing as a sample rate on a vinyl record.

a vinyl recording can be more accurate than a recording on a compact disk, a digital recording is not capturing the complete sound wave. It is approximating it with a series of steps (the sample rate).

http://electronics.howstuffworks.com/question487.htm

as for recording, you should be archiving your vinyl recordings at the highest native sample rate your interface can sample at.

44.1/24bit or 48/24bit or 96/24bit


Wait a sec: where does it end? There are cards capable of 24/192 now for a little over 100USD?

What about the 'Nyquist-Shannon theorem' as explained by oretez?

In doing my homework, the only reason I've been able to find for going higher than 48khz is to eliminate the pops and clicks on older vinyls when you remaster later, but since my buddies LPS are all new there seems to be little need for this. I don't even know that the explanation is correct, it's just the best one I could find.



Store the archives as an unprocessed and uncompressed, raw, high resolution digital wave format on DVD.


Sorry for all the questions (though in my defense isn't that what these forums are for ;) ), but why don't you recommend using a lossless codec to at least halve the size, more or less?


it is always best to process from a high resolution digital wave to a compressed truncated format like MP3.

cheers!

I guess that is your answer to the Nyquist-Shannon theorem?

Thanks for the response!

[Viracocha]

Vinyl doesn't have a sample rate no, maybe it's confusing because we're talking about frequencies.

Let me put it this way - your ears can detect frequencies up to say 20KHz. That doesn't mean you then have a sample rate of 20KHz :)

Don't record at a 96KHz sample rate. If the source contains no frequencies above 22KHz, then recording at a sample rate that captures up to 48KHz is obviously pointless. It's like using a 5 litre jug to store one litre of milk. There is no benefit to be had there.

If you had a source that contained frequencies up to 48KHz, then it would make sense, if you wanted to perfectly capture the source.
But then, the pragmatists would argue, that if you can't even hear the frequencies above 20KHz in the first place, it is still pointless.

So a bit of theory and real world there. 24 bit at 44.1KHz is fine for recording as dc has affirmed, in fact, as vinyl only has a dynamic range of about 70dB, you could arguably use 16 bit (96dB dynamic range) and get away with it no probs, but using 24 bit is still a benefit from a processing point of view, post recording.

There are no converters in existence, enthusiast or high end, that can either record at or output 32 bit sound.
When we talk about 32 bit, it's just referring to the processing done in the digital domain after the recording has taken place.
The more bits you have to play with, the more steps of detail there are that can be represented, and in the case of processing as here, that can be manipulated.
Every single gain (volume) change, any eqing, even panning, in fact any change whatsoever that you make to that original file, those are all processes that are taking place digitally, and every single one of them is basically just a mathematical calculation, so the more precise that calculation, the truer the result, and this is where higher bit depths come in to play, they basically allow the maths to be more precise.

Another analogy. If I give you a complex maths problem to solve, which involves numbers with 100 decimal places, but your calculator can only break down numbers to 10 decimal places, then your solution is going to be compromised, you will have to round off numbers.
If however your calculator can break down numbers to 100 decimal places, then you are going to get an accurate result.

So it is with bit depth. For every bit increase, the computer gets more levels of detail to work with, and it can produce a more accurate result.

You can only record at 24 bit, that is the current limitation. But in the software, with all these complex calculations, the resulting numbers may well run into thousands of decimal places, so to accommodate this, it's as if the computer is doing the extra maths in its head (which is what is meant by 32 bit floating point basically) before coming up with a result that can then be accurately portrayed in the 24 bit world.

This is very conceptualised so it's not a scientific explanation but hopefully you catch my drift!
One last analogy to ram the point home:

I give you a pen and a piece of paper, and tell you that you are not allowed to write down any number that exceeds 100 (24 bit in this example).
If I then give you a complex equation to write down and solve, and none of the numbers I give you exceed 100, then you can do it all on paper (staying within the 24 bit realm).
But if I give you another equation, and some of the numbers exceed 100, you will have to do as many parts of the equation as necessary in your head (32 bit floating point realm) until you get to a number below 100 that can be written down.
That's basically the concept.
So you can't physically output 32 bit sound, but 32 bit maths will help to accurately shape the 24 bit sound that you do hear, via processing and the mix engine.

[AndyH]

Signal level is easily seen while recording via the VU meter displays in any decent recording application. This is a good one, freeware for recording, shareware for its primary intention of cleaning up the recording (remove clicks & pops, etc.).
http://www.delback.co.uk/wavrep/

Also, the Audiophile 2496 has a good control panel display that is easy to use. This soundcard can generally now be purchased for <$100 since more advanced models have been developed. Its recording/playback quality is significantly better than anything you can get off vinyl, however, so there is no reason to go after more expensive cards.

Level too high (or hot) means exceeding 0dBfs. Unlike with analogue recording, there is no limiting or compression in the mechanism. Exceed 0dBfs and the input never makes it past the converter, you just get a flat, nasty clip. However, the earlier comment that someone made about not exceeding -6dBfs does not apply to any professional or semi-pro card, only to things like soundblasters. Good cards are completely flat until they clip when exceeding 0dBfs.

A few sample recordings will show you what to expect from your setup, whatever that turns out to be. If you don’t have a mixer or line-level preamp, don’t worry about it until you know you need one. I have not run into clipping once on the 600 albums I’ve recorded, but some people find heavy clipping from most LPs. As I said, it depends upon the cartridge and phono preamp.

You are saying the Denon receiver does not have a tape out? That is very uncommon. And unfortunate.

My comment about emotionalism refers to the people who can’t stand to record unless they are getting as near 0dBfs as possible. They think they are going to get better results if they don’t “waste bits,” but they are just making the task harder for themselves. Vinyl does not have the resolution for this to matter. Maximum peaks at-20dBfs to -18dBfs will give you as good a recording as peaks at -1dBfs.

A few albums I’ve recorded had some peaks just under 0dB, which is to say my setup isn’t lacking in signal strength; I’ve just been fortunate that nothing has been higher. The cleanest unmodulated sections, such as between music tracks, and a few totally unmodulated tracks on some test records, have a recorded noise level of -56dBfs, or higher, often not much better than -50dBfs. (My system [TT into soundcard, everything ready to go] noise level is -80dBfs, so it does not influence the LP noise.)

You can clean up the recording with filters and noise reduction, but the normally quoted figure of 60dB dynamic range for LPs is realistic. 12 bits is 72dB, 16 bits is 96dB. You can record at 24 bit if you feel compelled, but there isn’t going to be anything but noise in the lower 10 or 12 bits. Also, there are no 32 bit cards. The very best 24 bit cards are too noisy in the lower 5 or 6 bits for those bits to matter. This is a limit of physics, not to be improved upon without cryogenic cooling.

While most professional recording and processing is at 32 or 64 bit these days (24 bit soundcard captured into a floating point format), the inherent noise of vinyl means that you can process (declick, noise reduction, etc.) the 16 bit recording at 16 bit and never notice the difference (vs. processing at 24 bit or floating point).

There can be LP content exceeding the bandwidth of a 44.1kHz sample rate (i.e. above 22050Hz), but nothing you can hear. While this is a bit controversial in some circles, it has been pretty well established by careful objective testing that no one can tell the difference between any high resolution DVD-A or SACD recording (i.e. 24 bit, 96 to 192kHz) and a 16/44.1 resampling of same. I’ve done more than a few LPs that had nothing significant above 8kHz, which could be readily captured with a sample rate of 16kHz if you really wanted to save disk space.

Any digital recording must be bandlimited (to 1/2 the sample rate), but within that limitation it is not, definitely not, true that there is any inaccuracy. Everything that was in the original analogue signal is in the playback, not just some pieces of it. You get no improvement by using a higher sample rate than that dictated by the highest frequencies you need to capture. Besides, the raw sample rate for 44.1kHz is 64X or 128X with any modern card, 2.8 to 5.6 MHz.

I don’t know if anyone was actually recommending against a lossless encoding, but there is nothing to lose by such compression. FLAC is a good bet as it is probably the most commonly used and most widely supported, but there are half a dozen or so that will save you about 50% on storage space and deliver perfect decodings.. There is a downside in playing from a compressed source if your computer is too burdened to have enough resources to do the decompression in real time, but that is the exception, not the rule, with modern computers.

When you listen to vinyl, unless it is at a pretty low volume control setting, there is some feedback from speakers to cartridge. You have to put up with that when listening unless you can keep the TT isolated in another room, but you don’t have to accept it while recording. Monitor with headphones or the lowest sound level from the speakers that lets you know it is running, and forget about the party.

[oretez]

Don't want to wax too philosophic here but most of your successive questions have either been addressed or simple pointers were provided to let you resolve the issues with cursorary research.

That said the three assumptions you suggest that I just accept (for the sake of discussion) are simply wrong (not wrong in my opinion just wrong). That simply means that no matter how you twist my, for example, response it simply does not map to your inquiry. Which is why I tried to suggest how logistics are influenced by your definition of 'hi def'. Trying to side step your assumptions even before you listed them.

Two examples, amplifications of info already presented. If you have 100 albums storing 'hi def' audio is not an issue but if the library is 2000 you are now in 3 terabyte range required for 24/96 storage. While one can currently pick up 3, 1 terabytes drives for less then my first 500 meg drive (the first dedicated solely for audio) it is still enough of an expense to at least question whether tracking at bit depth & sampling frequency a little less overkill might let you spend more on a preamp with better power transformer, better isolation, less likely to interfere with the initial transfer!

OK, while theoretically a 24 bit card can provide 144 dB signal to noise (S/N) I've never worked with a real world system that even approaches that. I have worked with some upscale convertors that hover around 120 dB, but most entry level audio (not sound) cards designed for music production tend to come in around 109 dB (though a number of those wheeze dramatically trying to produce those results in real world context). AndyH suggests that you don't sweat the dynamics, leave yourself plenty of headroom. I don't recall his specifics but say try to keep normal peaks below 12 dB. You have just effectively moved your nifty hi def recording into 16 bit territory! So? Again if you have 100 albums you can listen through once, watching meters, making notes. On the second pass you can either set gain to maximum permissible for that tune or ride the faders based on your notes. If you have 2000 albums you are probably better off opting for the 9-12 dB of headroom, batch normalizing to -.3 dB after the capture.

In other words with careful attention you can achieve the same dynamics with a 16 bit recording that a more casual approach produces with a 24 bit card. Which is a better choice might be determined by how you value your time.

There is no magic one set of procedures fits all, produces 'hi def' audio . . . In point of fact your decisions will define what is, or is not, hi def. For You. And in point of fact what ever you do, what ever the original LP dynamics might be you will lose 6 dB in the original transfer. Which is another reason why, no matter what AndyH says, or he finds to be acceptable for his listening pleasure, getting the maximum dynamic range in the initial transfer is so important. In the digital domain you don't have to lose anything after that,. But anything you add after that is not part of the original source.

There are things you can focus on that will have measurable impact on your project: anything in the analog realm: condition of records, stylus, tone arm tracking, correct isolation of preamp power. Deciding that you are going to take the time to listen and monitor before you capture then setting gain for maximum dynamics during capture. I was attempting to side step the 'I really can hear cat sounds' argument by pointing out that pretty entry level physics suggests that you could not reproduce those frequencies even if you could hear them! So worrying about maintaining (for life) 3 terabytes of storage gets you into Emperor's new clothes territory pretty quickly.

But . . . There is no possible agreed on definition for 'hi def' audio because everyone will hear what they want to hear. Because I think those extra zero's packed into a 96 kb file were mostly meant to pack a marketing directors bank account and will be useful roughly when cats learn to speak does not mean that there is not a chance that in a couple of years you will be really, really glad you tracked 24/96 . . . Stranger things have happened.

[sabianq]

it sounds better,
it doesn't sound better
Truth, cause we can sort of tell.
I know I can pick out a live band playing over any resolution playback anytime.
can i hear over 20k? hel no, i can barely hear at all being a sound guy.
but i can absolute pick out live over recorded.


that is a moot point.^^^^^^^^^^^^^^

we use the high resolution formats because when processing them, they lose quality.

generational is bad for recordings,
you cant ever make better
you can only make worse
start at the best
always


everything single thing you do to a sample degrades it.
encoding, adding digital filters, anything you do to the sample cuts it up and changes that data.

when you start with a the highest resolution and you process it, it will sound better in the end.

recording to am mp3 is fine if you are in the field and you do not plan on processing it or if it is just a speech for reference.

but if you plan on restoring it or recording an original piece of art,
then you should have an archived original that is at the highest bit depth and resolution you can get, not because it sounds better, but because it will keep its integrity through years of future processing.

the higher the quality, the deeper you can process it while still letting it be the same.



:cool:

[AndyH]

That’s a hypothesis. What is the evidence?

A higher bit depth makes a difference in quantization error. This is a simple mathematical concept and it is objectively factual. That it is ever audible must be established empirically. You must also establish that it isn’t simply dithering vs non-dithering of 16 bit audio that makes the difference, i.e that the dither itself is audible.

The mathematics of what dither does isn’t too hard. There is such a thing as self dither. LP recordings self dither because the inherent broadband noise is so high, added dither is irrelevant. Find a sample where dither vs non-dither makes an audible difference from processing a 16 bit LP transfer at 16 bits. I’ve gone through this with professionals who told me I’m ignorant and dumb, but none could find anything to back up their claims.

Dither vs non-dither of very low noise 16 bit recordings can be audible, although it is easier to establish with generated test tones than it is with real music. However, the dither itself, although it eliminates the quantization distortion, can be quite audible if done multiple times for multiple transforms. This is a good reason for using a greater bit depth for processing, particularly a floating point format. The quantization noise/distortion remains too low to ever hear. No dither is necessary until, and unless, the bit depth is reduced for the final product.

That a higher sample rate can make an audible difference in the final product is much more speculative. It seems unlikely; I doubt you can find a sound theoretical basis for the claim. It would have to be established empirically.

The only valid process I can think of off hand is to make a high sample rate recording (recordings, for multiple tracks), resample (properly, with good software) to 44.1kHz, then process both in exactly the same way. If, and only if, people can differentiate the two in correctly done ABX tests do you have a basis for your hypothesis.

ABX testing is also the only way to tell if 16 bit vs 24/32 bit can make a difference when recording and processing from LP.

[sabianq]

http://www.masterworksound.com/Signal%20Degradation.pdf

[AndyH]

That self serving advertisement for MasterWork Sound Studios is about par for advertising. First let us set aside as irrelevant everything he wrote that doesn't deal with "DSP" processing. Some is more or less true, some is disputable, but none of it has anything to do with bit depth or sample rate.

He equates data change with degradation. Degradation is a highly charged subjective word. It implies that something quite bad happens. On this basis, that change equals degradation, one EQ transform, because of the EQ changes themselves, even if done to a thousand digits of precision, so vastly "degrades" the audio that none of the bit depth or sample rate results could possible matter in comparison.

I did not argue anything against the facts related to processing at different bit depths, the mathematical and objectively measurable differences thereof. I did say that often enough, and especially when dealing with very noisy material such as an LP transfer, there is no "sonic destruction," or any audible difference, if sticking to 16 bit instead of using a greater bit depth. Maybe you could find an ABXable sample where this is not true; I don't think anyone else has managed.

Anyway, the paper states "This unavoidably destructive component in DSP can be largely explained through a discussion of bit depth and word length." This is true about whatever "destruction" may occur. It has nothing to do with sample rate. Nothing I wrote argues with that except the "destructive" part. With the right material it does make a noticeable difference, but his "Practically speaking, every DSP operation yields some sonic degradation" is certainly nonsense when working in floating point and often not true for integer operations..

He is again effectively equating any data change with something bad, regardless of the fact that its result is inaudible. On this basis, the basic DSP operation itself is so relatively terrible it should not even be contemplated. Aside from those change one is deliberately making, the result of which is intended to be, and generally is, beneficial, the changes are just not "destructive" in the sense of causing harm. Quantization errors in floating point are not ever audible.

He makes a statement that many audio engineers don't understand the value of working at a greater bit depth or the dithering that should be done at the end. I have no evidence one way or the other, but I see this as improbable and merely an effort to attract customers to his business, away from competitors.

He then makes claims that great destruction occurs when converting down to 16 bit at the end of a project. He says that the ambience is lost along with those lesser significant bits. The Moran and Meyer paper, "Audibility of a CD-Standard A/D/A Loop Inserted Into High-Resolution Audio Playback" put the lie to that. That information should be well know by now, but of course it is quite inconvenient for those who have nothing more useful to sell.

He says several things about working with higher sample rates in the next to the last paragraph on page 3, and again near the very end of the document . He does not attempt to substantiate his claims. This is exactly what I wrote in my last post: If there is any truth to it, it can be objectively verified with double blind testing. Maybe it can, but I suspect it is snake oil, without objective merit. I don't think anyone has done such a study with positive results.

One last error about distribution formats, in particular mp3, although it isn't really relevant to the basic topic here. Besides making an inaccurate statement about the mathematics of the encoding, he says the result is "totally compromised" and "sounds inferior" even to the untrained ear. Mp3 can be done poorly, but it can also fulfill its intent rather well. Properly done mp3, easily accomplished, is transparent for most music to most people. By most I mean the high 90s percent. It cannot be differentiated from the original in double blind testing.

[sabianq]

Bob Katz shares this position.

question: Are 24 Bit and higher sampling rates really better than 16/44.1?

answer: Definitely... at least for origination. Maybe not that audible for destination. What I mean is that the higher the resolution of your source, the more generations of calculations you can go through without adding perceptible distortion or obvious loss of resolution.

question: But how does the 24/* sound when downsampled and requantized to 16/44.1? Does it really sound better than recording done at 16/44.1?

Answer: It can. Since you have choice over the dither. And if you have to perform any calculations (e.g. EQ or other processing), then you must dither at the end, and you should work at the longer wordlength to reduce cumulative degradation. It's one thing to say that you cannot hear a loss in one generation, but what about cumulative losses....

http://www.digido.com/faq/23-W/111.html

[sabianq]

all i was suggesting was that the higher resolution the original recording is, the better it will holdup through Digital signal processing.

nothing more.

test.

make a 24/96 recording of a piano.
make a copy and encode it to MP3.

then add EQ, compression, limitiation, reverb etc... equally to both the 24/96 and the mp3 samples.

it will be apparent which sample will hold upto mutiple processing better.

[sabianq]

Digital generation loss can happen in a digital recording.

http://en.wikipedia.org/wiki/Generation_loss
Techniques that cause generation loss in digital systems:

Transcoding – converting between lossy formats, be it decoding and re-encoding to the same format, between different formats, or between different bitrates of the same format – causes generation loss....

Generation loss caused by lossy compression can be made worse if the parameters used are not consistent across generations. For example, with JPEG, changing the quality setting will cause different quantization constants to be used, causing a lot of extra loss. Shifting the data in ways that cause a mismatch of splits (for example image blocks, video keyframes) in each generation due to editing can have similar effects.

Digital resampling, picture scaling, and other DSP techniques can also introduce artifacts each time they are used...


the same is true for digital audio.
the longer the wordlength, the better it will hold up to digital processing.

[FuriousGeorge]

Thanks for all the info guys. I'm taking it all in.

I think the answer that I got, which wasn't exactly the answer I sought though may have been better, is that a $100 phono-preamp and a $100 sound card should get the source to the final format with minimal/negligible degradation beyond what is required by the laws of physics.

Of course, we still need to be sure the original vinyl media is clean and in good shape, or the point is moot.

As to the bit-depth and rate, I've been doing some homework, and combined with the wealth of information I get from here, I've learned some interesting things:

-Vinyl may sound better, subjectively, to anyone, and there is really no right or wrong as to that debate, but, objectively speaking, digital PCM sound delivers a higher fidelity to the original source.

-No true scientific research has ever established that even the most 'audiophilic' amongst us can discern 16/44.1 from 24/96

-Even still, starting with a higher bit-depth/sample rate may result in better final products when converting to lower bit-depth / sample rates or remastering.

-If the final destination is more often than not cd media, I may benefit from sampling at 88.2 vs 96 to eliminate rounding errors when I down sample (right?)


But . . . There is no possible agreed on definition for 'hi def' audio because everyone will hear what they want to hear. Because I think those extra zero's packed into a 96 kb file were mostly meant to pack a marketing directors bank account and will be useful roughly when cats learn to speak does not mean that there is not a chance that in a couple of years you will be really, really glad you tracked 24/96 . . . Stranger things have happened.

-Then again, maybe 96 would be a better choice... ugh...

How am I doing so far? As always your thoughts, be they in agreement or not, are much appreciated.

[sabianq]

here is the thing.

your vinyl record is an analogue wave form carved into a disk of plastic.
while it is a truer representation of the live sample, we as humans cant tell.
however, a vinyl record can in reality be far worse than even a crapy MP3.

as the record plays, you get harmonics of the needle actually scraping the wall of the grove in the record. this can be really bad.
little tiny dust particles make the needle jump around causing pops and clips (the needle literally can jump off its track and skip portions of the sample).
the vinyl playback is full of noise by nature being analogue, the actual playback needle is susceptible to the laws of motion, every thing on the record that the needle touches will be sent to the output as sound.

(i would be willing to bet that with a sufficient bit depth and sample rate one could actually map the shape and texture of the groove wall)

^^^^^^^^^^^^^^^^^^^^^^^^^^ this is an application that ultrahigh bit depths may be able to realize.

anyway back on topic.
so in general, the record is actually a really bad source.
this is the number one reason to use the highest bit death available.

because, the higher your bit depth/resolution the more room you will have to restore the album.

you will be able to see the pops and skips better, adding filters, compressors, limiters and noise reductions will take better. The final encoded copy (Mp3, FLAC, OGG etc..) will be of a better quality.

also, you will have an archival grade master (raw, uncompressed and process free) original that encoded copies are made from...

different and more modern restoration software filters are being built all of the time.
some filters don't work on mp3 very well. even the most expensive and best mastering and restoration filters in the world do better with higher resolution master files.

you have heard the saying bout polishing ----.

[sabianq]

hey andy, i am not arguing here that there is an audible difference between 16bit and 24bit sample rates,

rather i an suggesting that the higher resolution the original master file is, the more you can do it before you do get to the point that the distortion from messing with it becomes audible.

and i am also arguing for archival interests.

i think it is silly to load up an ipod with HD audio wavefiles, they take up space.
however, the better the master the better the mp3 meaning that lower bit rates on mp3 sound better when made from higher resolution masters...

so you can actually have more songs fit on your ipod because you can make smaller files that sound good.

[FuriousGeorge]

Thanks again for the responses. You make some good points. Hard drive space is really cheap these days. It sounds like I have nothing to lose sampling at > 16/44.1, besides some time and bytes per LP, and I might even have something to gain when it comes time to remaster / resample / convert later (though not in terms of subjective audio quality of recording...)

What about the rounding errors? My buddy mostly listens to CDs when not listening to Vinyl, so (unless I can convince him to stop being a tool and learn to make a playlist) wouldn't he possibly be better served with samples at 88.2? Wouldn't that eliminate any issues resampling to 44.1 later?

Also, can you tell me where the point of diminishing returns is, in terms of $, when it comes to the phono preamp? It seems that no one takes issue with the $100 figure I quoted above. Obviously, when you are dealing with analog some degradation is inevitable, but at what price range does that avoidable degradation become negligible, even to discriminating consumers such as yourselves? A recommendation of a specif preamp would be appreciated, or could you tell me what spec I need to look for when shopping?

here is the thing.

your vinyl record is an analogue wave form carved into a disk of plastic.
while it is a truer representation of the live sample, we as humans cant tell.

Far be it from me to argue with anyone on these forums, but:
Everything I've read says that even 16/44.1 digital is an objectively higher fidelity (truer?) representation of the original sound.

On the other hand, subjectively speaking, many people prefer the sound of Vinyl, but, from what I've read, that is not because it is a more accurate medium, but because it adds some 'nice' or 'warm' or whatever distortion.

I take it you disagree with those statements?

[sabianq]

get a good two channel usb audio interface 100-200 bucks

honestly, all of the actual hardware today is about the same quality.
it comes down to drivers and better interfaces will have better drivers for the hardware.
presonus has a good reputation for good drivers with a good piece of hardware.

http://www.sweetwater.com/store/detail/AudioBoxUSB/

here is a lexicon Lambda demo for 130 bucks
http://www.sweetwater.com/store/detail/Lambdad/

[sabianq]

Thanks again for the responses. You make some good points. Hard drive space is really cheap these days, and it sounds like I have nothing to lose sampling at > 16/44.1...

i suggest storing archives on cheap thumb drives (apparently the data retention on flash if it is not written on more than once is a really really long time)

What about the rounding errors? My buddy mostly listens to CDs when not listening to Vinyl, so (unless I can convince him to stop being a tool and learn to make a playlist) wouldn't he possibly be better served with samples at 88.2? Wouldn't that eliminate any issues re sampling to 44.1 later?
honestly, it really doesn't matter. as long as you like the final product.
I am just saying that if you need to really restore an art work, and need to really process it to make it listenable, a higher bit depth to start will give you better final results.

however, the Native bit-depth of the hardware will produce less errors.
(I always record at the native bit-depth of the cards or interfaces A/D hardware.)



as a master file, the best sample rate is IMO always the better option because anytime you re sample, encode or process, you lose quality.
and it is easier to make smaller and better sounding compressed files.

plus it is nice to have masters around if you want to do something different in the future.





Far be it from me to argue with anyone on these forums, but everything I've read says that 16/44.1 digital is an objectively 'truer' (higher fidelity?) representation of the original sound.

not higher fidelity, it is merely an analogue wave, meaning only that the whole thing is there, the whole wave, it is not a whole bunch of little slices of information based on time and placed in sequence to represent an audio wave. an analogue wave cut in a piece of plastic is in effect a mirror of the wave the sound system heard. (every, noise, hum, pop click, ambient sound, vibration etc.. present in the system when the master was cut, is in the final vinyl recording) the whole wave, a continuous, uninterrupted actual sound wave cut into a medium by shear force.

There are no time slices of the wave. even the best 1 bit 5 megahertz recordings still leave a bunch of the audio spectrum "unheard" (unrecorded), because in "theory", you can cut up an analogue waveform into time slices as small as a planks scale meaning having a sample rate of 10^18 samples per second.

a whole analogue wave by nature has far more information in it than any bit depth digital sample.
whether or not we can hear it or the extra information makes it sound better seems to be a personal matter alone.



On the other hand, subjectively speaking, many people prefer the sound of Vinyl, but, from what I've read, that is not because it is a more accurate medium, but because it adds some 'nice' or 'warm' or whatever distortion.

I take it you disagree with those statements?

i personally think all vinyl is noisy and sandy.

i like the quietness (is that a word?) of digital.


but i absolutely prefer listening to someone playing a violin or other instrument in front of me.

[sabianq]

if anyone is interested, here is a great article on the mechanics of a vinyl record.

http://www.badenhausen.com/VSR_History.htm

Already in 1954 high requirements for audio reproduction were discussed in the USA for improving the existing High Fidelity standard:
The Radio Designer's Handbook (Langford- Smith) was referring to a frequency response of 30 Hz to 20 kHz at a linearity required ± 0,1 dB! For this ambitious aim to better performance, the distortion factor was proposed to be less than 0.5 % (THD) at an S/N ratio of 62 dB.

and apparently the term loo/m.;l(+l;oi

[FuriousGeorge]

Thanks sabianq. The explanation makes a lot of sense, and according to my spell-checker quietness is a word! ;)

We've found a mixer: A Delphi PMX-02
Also, for the record, the turntable is a Newmark TT200

Please don't tell me that this is crap :)

That leaves us needing the soundcard (think we'll go with the suggested M-Audio 2496), a vinyl cleaner (discwasher 1006), and some decent plain old rca cables.

[sabianq]

i am sure it is fine,
like i said,
i am not one for viny, my dad had a huge collection and a nice linear turn table i always remember being able to hear the "hisssss and crackle" when he played his albums.

when he got his first cd player,
the music got noticeably better.

i honestly thing there is no issue with your equipment list.

good luck
cheers!

[AndyH]

Something invalid has been inserted. Perhaps it is my fault for mentioning the topic and not making something clear. MP3, and any other “lossy” perceptual encoding, is an end format only. No one does serious work using them as a source or storage format. That isn’t their purpose, and any comparison made between the results of working on them vs working on an uncompressed format are not meaningful. That reference on wipikedia was about such encoding; it is not relevant to the discussion here.

I just read an argument about the superiority of “higher resolution formats” that used as example a very low level sine wave as some sort of proof. In 16 bit, at very low level (as in down around -96dB) it sound very bad while the 24 bit version sounds correct. One must, of course, be listening at a very high volume setting, fatally high for any real music.

Duh, of course an undithered low level 16 bit tone, used in the supposed proof, is quite distorted. Properly dithered, however, it is no different than the 24 bit tone at the same -96dB, except that you will probably also hear the hiss of some dither when you turn up the volume as far as necessary to hear the tone (at normal listening levels, good dither is not audible).

The above is representative of some of the things being written here, confusing some issues. Nothing I’ve written argues against the superiority of 24 bit. or better yet, floating point, over 16 bit for processing, and in many cases, recording. I’ve only said that it doesn’t matter for a recording from vinyl because of the noisy (self dithered) nature of the source. The 16 bit quantization errors are real, they are much larger than in 24 bit, but no one is going to be able to tell by listening.

I’m not even saying you should not record or process vinyl transfers at the higher bit depth -- just in case, ‘cause you never know, there might be something on some LP--. Just don’t fool yourself that you are actually getting something audibly better.

Higher sampling rates are a different topic. There is no necessary relationship between sample rate and bit depth, so try to keep the arguments separate. There is no evidence, only some unsubstantiated opinions, to suggest that higher sample rates produce anything superior for audio.

While some people find it inconvenient to accept the evidence about perception, and others find it emotionally unacceptable, the facts have been available for many decades. People’s beliefs and expectations strongly, even overwhelmingly, influence their perceptions. By now, millions of people have experienced the fact that some differences in audio they believed to be marked and obvious don’t really exist (experienced through participating in ABX tests). When subjected to proper double blind testing, the only thing people have to go on is what they hear. Without knowing which they are listening to, they can’t tell which is which.

All such objective evidence to date says that higher sample rates are not audible. This hasn’t been tested, as far as I know, in regard to using higher sample rates during the mixing, mastering process. However, while the facts explain why a greater bit depth produces better results mathematically (even is it isn’t audible, it is unquestionably more accurate) there are no such facts to support the idea that a higher sample rate will produce anything better.

A higher sample rate is not more accurate nor does it retain more accuracy through processing. When you apply a transform (a DSP process) to higher sample rate audio, you just apply it to more samples at a time. Maybe using a higher sample rate will, in some as yet unknown way, make for a better final product, but there isn’t any actual evidence to that effect nor any good reason to believe so.

There are no time slices of the wave. even the best 1 bit 5 megahertz recordings still leave a bunch of the audio spectrum "unheard" (unrecorded), because in "theory", you can cut up an analogue waveform into time slices as small as a planks scale meaning having a sample rate of 10^18 samples per second.

a whole analogue wave by nature has far more information in it than any bit depth digital sample.
whether or not we can hear it or the extra information makes it sound better seems to be a personal matter alone.

This is just a lack of understanding about digital signal processing. It isn’t true. This is a matter of physics, not faith. The only limitation is frequency bandwidth. You can only capture frequencies up to f/2, whatever sampling rate f happens to be. Within the Nyquist limit, all information is captured.

[AndyH]

Consider your high school geometry. How many points does it take to define a straight line? a triangle? a circle? If you use more points, are they defined more accurately, or are they still the same?

[oldschool]

All,

I have read your dialog with great interest, and it's probably one of the most detailed and interesting posts on this topic (recording from vinyl to PC) - at least from what I've seen, and I've googled for a couple of days now - so thanks and kudos!

First off, FuriousGeorge, I don't think anyone would classify the Newmark turntable as crap, but I doubt it will be considered high-end. From what I can tell, it leans more towards the DJ crowd, and priced at $160-200 (without cartridge) it's fairly entry-level (or even "sub-entry" level in the more audiophile world). The cartridge will make a difference, too, of course. Don't know anything about the mixer, but again it sounds like DJ gear, which typically is more geared towards features/useability and durability than high-end sound quality. But it's probably a good pairing with the turntable, and you may well get very decent results for your use.

Given the audio equipment information, I would tend to agree with earlier comments that you should perhaps not fret about the digital stage of the recording process too much (at least in terms of bits and sample rate) as long as it's a clean, distortion free signal path. Caveat here is that I am no expert of digital processing (even after reading all these posts :) ), and it's very possible that a higher "resolution" recording has benefits down the line. But not in order to capture information coming out of the mixer, as it will be well within 16 bit / 44.1Khz - if I have understood the technical points made earlier correctly. Finally, I can also second the thought on ambient sound during the vinyl recording - I once captured a dog bark when recording an album which could very clearly be heard in the background on the recording afterwards.

Either way, the best of luck with your project!


Now on to a question of my own - very much on the same topic. I too am looking to capture my albums (maybe ~100 that are worth recording). This was actually triggered by a christmas present from my wife - one of those USB turntables you just plug into your PC - but my thinking was that the sound quality couldn't possibly come close to that of my existing turntable and cartridge, so I decided to return it and spend the money on other gear that will work with what I already have (yes, the wife was very supportive of this move ;) ).

I have decided to definitely use my turntable (Rega Planar 3 with the Rega RB 300 arm and Audio Technica OC-9 MC cartridge - and I just upgraded the motor). My phono pre-amp sits in a NAD 1000 pre-amplifier, again I am intending to use this as it's supposed to be quite good. Finally, I also own an older DAT deck, a Denon DTR-2000 (from around 1990). And there is the main source of my dilemma: Should I pass the analog signal from the phono pre-amp through my old DAT player's A/D converter and run a digital signal into a sound card on my PC, or should I splurge on a more expensive sound card that might do a better A/D job than the DAT player?

Some specs, if helpful (I know sound quality is not numbers, so this may be redundant):

Cartridge:
- Dual Moving Coil
- Frequency response: 15-50,000Hz
- Channel separation: 30dB

Phono stage (moving coil setting):
- S/N ratio (A-weighted with cartridge connected): 76dB
- THD: <0.04%
- RIAA accuracy: +/-0.5dB
- ground connector

A/D converter in DAT player:
- Sampling: 32/44.1/48kHz, 16 bits
- S/N ratio: 90dB
- Total dynamic range: 90dB


So I'm seeing 2 main options here:

1) Record from turntable via phono stage and DAT player into a digital input on a sound card.
Is sound card quality important in this case, since it's not really doing any conversion, just receiving the digital signals so I can record them as files on my hard drive? I currently only have an on-board audio chip with no digital input, so I'd have to buy a card regardless (or some sort of device allowing me to capture either coaxial or optical input). Would a cheap-ish card, such as the Diamond XS71 DDL or Turtle Beach Montego DDL, both with optical inputs and running ~$50, do the trick or might they distort the DAT's signals on the way to my hard drive?

2) Buy a higher quality sound card with solid A/D converter and record straight from the phono stage to the sound card's analog input (and sell my DAT player on eBay, since it's been collecting dust for well over a decade now, and maybe pocket $100 if I'm lucky)
In this case, I understand M-Audio's Audiophile 2496 is a good choice, and at least on paper well exceeds my DAT's capabilities at up to 96Khz sample rate, 24 bits, and S/N ratio of 100dB for the A/D converter.
The other high quality option I've looked at for home theater capabilities in addition to a good A/D converter is the Asus Xonar D2 (up to 192Khz sample rate, 24-bit, 118dB S/N, 7.1 Dolby Digital and DTS and all that stuff). This card also looks like it's shielded, which may or may not provide real benefit.

If I'm actually able to sell the DAT player it will effectively pay for the sound card upgrade from a ~$50 card to either the ~$90 M-Audio or the $170 Asus, which says option 2 is preferable (also less gear and cables to deal with).

But am I missing anything in terms of advantages of the DAT solution? Are there going to be quality differences in favor of using an "outboard" A/D conversion, either because it's away from the computer, or simply because the DAT's A/D converter is likely to be of better quality (outside of pure numerical specs) given the hardware itself?


I appreciate this may be a subjective call, but would appreciate any input/suggestions.

As a final note, I must say that subjectively I've never been able to tell much of a difference between a CD recorded on the DAT player and the source CD, even when the CD has been recorded analogally (is that a word?), i.e. analog output from CD player recorded into the analog input of the DAT player. Would suggest to me that the D/A and A/D converters are doing their jobs, and that perhaps for most practical purposes there is not too much to worry about when it comes to the A/D conversion process? Here is an older article on DAT, but it also covers a good deal of "digital theory" as well - some sections are quite good and still valid today: http://www.saddleback.edu/faculty/mschiffelbe/ca118/handouts/dat.htm


Many thanks!

PS! FuriousGeorge, sorry for hi-jacking your thread... :D

[AndyH]

Your suggestion may give you quite acceptable results, using the DAT deck, but you may be more happy in the long run to have a soundcard that is less restrictive for other projects. Multimedia/gaming cards record everything at 48kHz, then hardware resample to the sample rate you specify (this is a Microsoft multimedia specification). The cards you mentioned probably have that restriction, though many attempt to mislead the potential buyer. Some of these cards will do a straight digital transfer but many resample there too. You are probably fine with the vinyl project if you work at 48kHz, them resample to 44.1 with software. Some such software is good, some isn’t so good.

However, once started, you may find an interest in other projects. The DAT deck is not going to be superior to decent PCI cards or other external modern cards. You will be restricted with those multimedia cards in terms of serious audio work. The previously mentioned Audiophile 2496 would be a good choice, and there are other professional/semi-pro cards that aren’t terribly expensive. Echo and Emu are quality sources, there are a fair number of others.

When the Audiophile 192 came out, the 2496 dropped from about $250, if I recall correctly, to $99. I have seen it for $70, but I have not paid attention to the market for quite some time.

[AndyH]

The DAT deck probably has the advantage of input level control. If so, this eliminates the need for a separate line level preamp, which you might or might not need when going from phono preamp to soundcard line level inputs.

[ecc83]

I have done quite a bit of this.

Iam NOT going to wade thru' this lot so please forgive duplication.

Yes, go 2496, not a lot of point going better tho' IMHO.

There is NO quality difference 16bit and 24bit and anyway who here has ever heard 24bit reproduction? But,

24bit(at 44.1kHz) has effectively an infinite noise floor so this is the process.

Set the average level on the software meters to -20dB or so, yes it will look mighty low, worry not. Ensure that music peaks get to -10 VERY infrequently and -6 is a no-no. This ensures that spikes do not as far as is possible drive the conveters into clipping.

Run the recording thru' a good vinyl click and noise suppressor program, an excellent one is in Sony Soundforge which you can download a demo of which gives you all the features including save for30 days. In the software up the volume to average neg 15 or so depending on DR of the content. Export as .wav and burn to taste. I use Nero (normalizing off) but there are many such.

Saw my sobriquet and "attenuator"? Eme and I will oblige.

[oretez]

it sounds better,

can i hear over 20k? hel no, i can barely hear at all being a sound guy.
but i can absolute pick out live over recorded.


that is a moot point.^^^^^^^^^^^^^^

we use the high resolution formats because when processing them, they lose quality.

generational is bad for recordings,
you cant ever make better
you can only make worse
start at the best
always


everything single thing you do to a sample degrades it.
encoding, adding digital filters, anything you do to the sample cuts it up and changes that data.

when you start with a the highest resolution and you process it, it will sound better in the end.


:cool:

responding to this is a bit off topic and by and large I do not disagree with spirit of main points ( reason to track at highest efficient (based on budget considerations not purely aesthetic ones) bit depth and resolution you can is not for subjective perception but for objective processing considerations . . . I vastly prefer live music and after 30 yr. of effort am frustrated by how poorly any & all technology reproduces 'live' sound. But many of the psycho acoustic reasons you can distinguish live from record, blind folded and half unconscious; has to do with variables completely divorced from specific capture technology)

but there are a couple of points I want readers to have additional perspective on

First, generally speaking, you don't listen to 'digital' . . . (you can and I have even without having installed a jack in the back of my head) you listen to 'analog' you hear analog . . . the 'digital' is encoded information that is then reproduced via analog gear . . . there are no 'holes' in the signal . . . if there are anomalies in signal process . . . you will be aware of how a 'how' would be represented. So the theoretic question is whether the encoding process accurate represents the original information. Nor can, in all likelihood, any representation be congruent with the original . . . original is 'bandlimited' in someway, no matter the technology.

the Nyquist-Shannon theorem, Whitaker-Shannon interpolation formula, Poisson summation formula, Shannon-Hartley theorem, & Hoelder (spl) exclusion principle might not have validity . . . it's possible . . . but if they do with in the bounds they establish the information they encode accurately represents the original

Understanding what it a digital representation of source actually is, is first step in being able to understand the triviality of most anti digital rants (and to be quite honest I am an 'analog' guy . . . material tracked to with tube mic pre's to 2 in tape (15 ips) sounds inherent 'musical' to me in a way that pristine digital does not (and quite frequently I still suggest & sell the idea of mastering from tape . . . dumping master digital back to tape, mastering from that) . . . that is never going to change, in part it is a function of when I grew up, what the source material was when I started listening critically . . .but many (if not all) of the 'musical' elements I hear in analog, that I find missing in digital are in effect distortions added or emphasized by the technology.

Nor does any one with industry credibility claim that even the 'best' analog systems offer a more accurate representation of original source

2nd, following from idea that any 'capture' by any technology introduces artifacts, and fails to be congruent with source, if one is working either as upscale archivest or as forensic technician then there are legal, contractual issues with digital degredation of original source the powerful tool 'digital' information storage offered was that while anything action 'changed' your data, those changes do not, necessarily, present degradations of the 'information' derived from the source.

That 'digital' (audio for this discussion but not limited to audio) captures 'information' not an 'analog' of the thing itself and that the ensuing information can be manipulated without inevitable 'degradation' of that information is at the heart of what digital 'is'. And as I say there is possibility that we have deluded ourselves concerning validity of Nyquist et al . . . until Pastuer we were absolutely sure that disease agents generated spontaneously .. . can you imagine the radical shift in thought process needed to accept that protection against histories greatest killers could be conferred by embracing and deliberately injecting them into our bloodstream?!

As I said at beginning I do not disagree with broad strokes, nor how it applies to the specific topic @ hand but 'digital' is not merely analogs short bus step child . . . it is a quantum shift in how information is processed and it's a shift that has been evolving for at least a couple of hundred years

with regard to audio we don't 'capture' audio we generate information about audio . .. .the most important variables in how that information represents the abstracted source is still the analog chain . . . which includes the room in which the pressure waves are generated, the ears and ultimately the nural process those waves impinge

to beat the dead horse dead to not 'get' the differences between 'analog' and 'information' means that gradually over time an individuals thinking will diverge from what is functional

[oretez]

with regard to old schools inquiry

In 1990 I would have had no issue with converters on the Dennon DTS 2000

but my experience suggests that there has been significant improvement in converters in even entry production audio (as opposed to 'sound') cards

my guess is that the card AndyH continues to recomend is enough 'better' then the DAT to be worth the effort to install

(whether you could hear the difference is of course is indeterminable)

certainly if you are happy with your analog signal path then pre amp out to audio card in is a simpler cleaner (with some caveats concerning computer power self noise and unfortunately even with s/pdif transfer you still face some of those same issues, plus you introduce potential master/slave clock issues between a consumer 20 yr. old DAT and current built to price consumer computer, theoretically the s/pdif carries clock info but your onboard card has to be able to slave to the Dennon ) then introducing the 20 year old Dennon

ultimately, for me and what I do (which is not restoration, old catalog transfer) it would be the quality, of the Dennon, of the filters at Nyquist frequency that would be my biggest concerns

I last upgraded by stereo DAT in 96 and it was the same year that the A/D card I got for home use (a Gina20 by the way) catgorically surpassed those available on the DAT's . . . even then, even at home I had a custom A/D to hard disk (by today's standard obscenely expensive with what now appears to have been a ridiculously short half life) but it was the off the rack impovement of the Echo card that surprised me and shifted my approach from custom to using off the rack components to construct DAWs

unfortunately matching an audio card to specific computer still has some pitfalls . . . the aging PCI entry cards by EMU & M-Audio are both readily available, relatively inexpensive and fairly safe bets (with regard to compatibility) I've been quite happy with all my Echo stuff (other then they tend to not play nice with other cards) But it does not hurt to check manufacturers web sites for known issues . . . early in selection process

[dcwave]

There is NO quality difference 16bit and 24bit and anyway who here has ever heard 24bit reproduction?

:eek: :confused:

[oldschool]

AndyH, ecc83, oretez - thanks for the feedback.

Consensus seems to be to ditch the old DAT and go for a card with good A/D converters and stable drivers. My shortlist right now is M-Audio Audiophile 2496 (~$85), the EMU 0404 PCI (~$90), and the Echo Mia ($130), all seem to have the right capabilities (and fits my budget).

I guess the only remaining consideration for me is whether or not to add multi-channel home theater capabilities for movie watching, since it looks like I'll be storing my album collection (and perhaps CD collection as well) on the PC and hence might keep it next to my hi-fi and TV. And since I don't have a home theater receiver yet, well, perhaps such a sound card is the way to go and just hook up regular stereo power amps (of which I have several idle ones) for the different channels. My reservation, though, is the possible quality sacrifice on the 2-channel recording side by going to a 7.1 channel card with all the associated gimmics. As mentioned in my first post, I've come across the ASUS Xonar D2 card, which has impressive specs and good reviews, and at ~$170 would be a reasonable premium over regular 2-channel in order to get a home theater setup. HT Omega also carry a few that on paper at least look compelling. Does anyone have experience with these and whether they still have A/D converters and recording capabilities on par with the first 3 cards mentioned above?

Thanks!

[ecc83]

Well ok Dave!

But what I meant was that there is no popular media/player for 24bit depth recordings.
Dave,

[AndyH]

I can’t say about in regard to very many particular multimedia/gaming cards such as those from ASUS, but the majority have converters that only work at 48kHz. When you record at a different sample rate, it sends the 48kHz through the onboard DSP chip to be resampled to your target rate. Same thing in reverse for playing other sample rates. Creative used to lie about this until they lost a lawsuit over it, now they just try to mislead potential buyers.

Here, perhaps the most important question is “how good is the resampling?”It has often been less than steller. To get an idea of what I mean, compare the Adobe Audition results, with proper filters, to Goldwave or Ableton Lite on the sweep frequency downsample. There are other programs that also generate extreme artifacts. With those kinds of cards you get the best results they are capable of by recording at 48kHz. Then, hopefully, you use good software to downsample to 44.1 if you want to make audio CDs.
http://src.infinitewave.ca/
The point being, is the audio work or the entertainment most important to you?

You should also be aware that these entertainment cards are not true multi-channel. They decode multiple channels multiplexed into a single file, they can not deal with multiple source files, one for each channel. This isn’t a problem for entertainment sources, such as movies, but is not a very useful approach if you want to work on your own multi-channel music.

You might want to consider the I/O defaults on the card. Home components generally use RCA jacks and plugs. The Audiophile 2496 does also. The Echo Mia, while quite good audio wise, uses 1/4 inch phone plugs. Conversion cables are not hard to come by but this difference isn’t always convenient. On the other hand, if you are mainly connecting to other professional equipment, the RCA plugs are the inconvenience. Many of the entertainment cards use 1/8 inch phone plugs, which I would call a real pain for any serious work.

[ecc83]

Andy H,

Totally agree, almost!

Yes the 2496 uses RCA's but they are gold plated and very solid with a decent plug. You are always going to be limited by the pci format.

In my case the RCA outs came up 1.5mtrs unbalanced to a pair of OEP 1:1 trannies and then out on TRS. Now they go 1.5mtr unbal'ed to a mixer and exit balanced.

Inputs go 1.5mtr to a TRS breakout box and 300mm away is a pair of OEP hi Q input transformers should I want to feed from tother side of the room.

Bottom line, sort your connectors to something rugged.

But I agree, 1/8"jack, lower than whale ***T!

Dave.

[oldschool]

Andy H et al,

Thanks for the feedback. While it appears (on paper) that some of the higher end entertainment cards may have a chance of coming close to the 2496 in sound quality (e.g. they claim 192KHz/24Bit for all inputs and outputs, 118db S/N, etc), I did decide on the 2496 from M-Audio in the end and got one from Amazon for $85. And yes, the 1/8 jacks of the entertainment/gaming cards did concern me, too - hard to take the hardware seriously when you have to plug in a "walkman" plug into it... But I guess it's the only way to fit 6-8 channels out and 2 in (plug digital in/out) on the back of a PCI card - without a breakout cable, at least, which none of these cards seem to take advantage of.

Installed the 2496 a couple of days ago, and while I haven't yet gotten around to do any vinyl (or other) recording yet, I hooked up a pair of electrostatic headphones via the analog out RCAs. And the card sounds fantastic! Heck, even my mp3s sound good through this card. This is definitely going to be good enough for my purposes, which is mainly home audio (not professional recording). I guess now I just need to go out and get a bigger hard drive for all my lossless recording.

Again, thanks for the guidance!

[ecc83]

Morning Oldschool,
So glad you are kitted out properly! The Delta series control panel takes a bit of getting ones head around so come back or PM me if you get stuck.

Recording large audio files is not what pc's were designed to do and there are several "tweaks" that improve the process immeasurably as well as keeping your temper at bay! Most important of these is to turn off Windows bleeps and bloops, they will screw your results rotten.

Over the course of this thread there has been much discussion of things digital, some good, some not so good and some just plain wrong! Get thee to www.soundonsound.com and register with their forum. Here you will get the unvarnished truth about bit depths and sample rates. You will also find THE articles FOR FREE about optimizing computers for sound.

Dave.

[oldschool]

Thanks, Dave. Agree that the control panel is a bit less than intuitive, but I think I've figured out what I need for now - at least I was able to get audio through the card and back to my headphones both from the analog input and the spdif-in (DAT player). Now I just need to master the recording software - I'm giving Audacity a try (primarily because it's free and it seems to come recommended for my use). Have registered over at soundonsound and will read up on PC recording there.

Thanks again!

Vidar