Lead Vocals Level Question (PC Multitracking)

[mark4man]

My primary drive & OS are beginning to crap out; & my sound card's hdwe effects are becoming unstable (that is...I think the two are related, but if not, I'm swapping drives anyway; & upgrading to XP.) I need to complete & upload one tune to my site prior to this task; & normally the method I would use to add presence to the lead vocal, is double-tracking. In trying to accomplish this now tho, with weird reverb, instead of the two duplicate vocal tracks adding to the gain, I just get clipping. So, my question is: Can I go with one good vocal track, drop the gain on everything else; & then normalize to get back to 0db, without loosing bit depth; & in turn audio quality, on the lower volume tracks? I have read that: the hotter the signal, the greater the bit depth; & hence the higher the quality of the sound...& conversely, lower quality sound for tracks recorded with a weaker gain. Thanks.

mark4man

[DAW-Freak]

Originally posted by mark4man:
I have read that: the hotter the signal, the greater the bit depth; & hence the higher the quality of the sound...& conversely, lower quality sound for tracks recorded with a weaker gain. Thanks.

mark4man [/B]

Where have you read that???
If I record "silence" at 24 bits it stays 24 bit. I would be more concerned about the signal to noise ratio.

[narcoman]

In a way DAW-freak he's absolutely right. If you record into a 24bit recording system and you're only getting levels halfway up you're not using all 24 bits. 24 bits is the max amount of bits available to describe the positions of your samples from AD conversion in the audio domain. I'm not sure, but it may even only be 23 bits and one bit for mantissa, someone correct me if i'm wrong..
If you normalise your silence (and lets suppose its card recorded silence, and not absolute black) then you will hear lo bit depth noise upon playback, so indeed if you record very low level signals then you will here a lower bit rate playback. However, as the 24 (or 23 ) bits in question equate to a base two scale then a signal at half the height does not mean its only 12 bit. A 24bit signal at half the height is essentially 23 bit. Half again is 22bit etc etc


cheers

[DAW-Freak]

Hi Narcoman,

So do you mean that if I record a very dynamic "instrument" like vocals in 24 bits the actual bitdepth will pump up and down like crazy, thus making the softer parts noisy and crappy sounding??? http://www.audioforums.com/forums/confused.gif

[narcoman]

in theory yes, but the reality is that you wont get down to your peaks only using two or three bits. Even if you go pretty quiet you'll still be using 18 or 19 bits. This is one of the reasons why most classical recordings are better served at 24bits rather than 16. There is a much larger dynamic range. The numbers to describe the amplitude of a signal in a 24bit wave are 2 to the power of 24, whereas in 16bit its 2 to the power of 16. I think in reality its one bit for neg/pos and the other 23 to describe the wave but the principal is the same. A 24bit number can be at maximum 16777216. A 23bit number can be 8388608 and so on. This means that to be a 23 bit wave then your peak must be half what the peak would be on a 24bit wave. This carries on all the way down, such that a 2bit wave can only half half the amplitude of a 3bit wave. With signal to noise ratios taken into account when your recording from an AD you will not get a 0bit or probably even a 1bit signal. I'd probalby say you wont get a 2bit either, but maybe people who know far more than me can correct me on all this. This is crawling back into half remembered education here !!

cheers

[narcoman]

Just to add, you'll still be storing it in a 24bit file , i'm not saying you have a dynamic bit sized file !! that would be a little mad.

[DAW-Freak]

Hello narcoman,
and thanks for the information.
guess i'm much wiser now... http://www.audioforums.com/forums/biggrin.gif

[narcoman]

ha,
Yeah i'm gonna try and change a few physics laws so we dont have to bother with all this....

[mark4man]

Guys,

Haven't had time to respond to DAW-Freak's original question...been busy of late, but it sounds like narcoman has a pretty good hold on the technical component. The following is from an online artistPRO Hard Disk Recording course, entitled "Gain Change/Normalizing", written by the great Bill Gibson:

>"We know that when the full amplitude is utilized in the digital waveform we’re using all of the available bits. In addition, we know that when only half the available amplitude is used in digital audio we’ve only really used half our available resolution, or bits. Therefore, it makes sense that adjusting the waveform gain to maximum, or normalizing, would be the right thing to do for the sake of audio integrity and efficient use of the available audio bits. In some instances, this theory holds ground; in others it doesn’t.

As the processor calculates gain changes from low-level signals, we run into problems. In the digital domain, low-level signals are best left that way. The resolution in very low-level signals is also very low, so when the gain is changed the result is simply a louder version of a low-level, low-resolution sound. Increasing the gain of poorly recorded audio does not increase the resolution. If you only use six bits of a 24-bit word and then turn up the level to occupy the space of a full 24-bit signal, the resulting audio doesn’t sound like 24 bits. It sounds like a loud 6-bit waveform.

The very best solution to this normalizing process is to record your original audio as close as possible to maximum amplitude. In this way, the only sounds recorded at very low levels are sounds meant to be there. In terms of waveform amplitude, the only way to optimize the clarity of your digital recordings is to use the full amount of available amplitude".<

mark4man



[This message has been edited by mark4man (edited 06-14-2002).]

[DAW-Freak]

Yeah, I'm glad that back in the analog days I learned to record my tracs with as hot as possible input levels so I don't have to worry too much about all this bit-stuff. The only thing that sucks with digital audio is that there is not really much headroom before you get digital clipping and that sounds UGLY. http://www.audioforums.com/forums/wink.gif

[narcoman]

hellooo again,

This B.G. quote is interesting. When you have a half amplitude signal actually only lose one power of two of the signal resolution. So at half amplitude you get max 23 bit sound. Its not as bad as it first looks, a half height waveform doesnt use 12bits !! Every time you lose half the amplitude ( i always nearly write amplitube), you lose a bit. So to get a six bit resolution signal from 24 A/D you need to be putting in a very low signal. The signal would have a max height of 2 to the power 6 (64) against the 24bit signal being 2 to the 24 (16777216). This means the amplitude height for a 24 bit signal is 262144 times as high ! or in other words dont worry about it !!! I think all the issues about reverb tails sounding slightly bitty are when you get down to 12 bit, which is still very quiet compared to a 24 bit signal, but not so with 16 bit.

cheers

[creativak]

I think that you need to use a little compression, maybe your vocals are so weak that you need to streng them up with a little dynamics tweaking, that's why it clips and it sounds weak at the same time...dynamics are the key.

By the way, you could get a cool vocal fx if you double track your vocals and spread them up a little bit in your stereo field, but always remember to check for mono compatibility if that's an important issue for you.

Cheers.