Voice Mapping?

[apegrass]

Greetings-
I'm told there is a way one individual can record a sentence, and using some software, the tone, pitch, and other specifics can be adjusted to match those of an exisitng recording?
Ex: we have an informational video which we paid a voice over guy tons of money to do for us several years ago- now we'd like to add 2 words to the video... but we'd like to be able to match the vo guy's everything... any suggestions?
Is this called voice mapping?
I've downloaded Adobe Audition trial software and have looked it over briefly... feeling a little lost at this point, any assistance is greatly appreciated.

Thank you.

[robertruetz]

I'm not aware of any software that can do this. That's why "voice guys" still have jobs. You can use pitch-shifting and EQ to alter your new voice, and try to get things close... but good luck getting an exact match. You'll have to do all the heavy lifting with your ears, there isn't any software/magic that will provide you with an exact match.

My advice, pay the guy to re-cut the sentence, otherwise you're throwing away the money you've already spent making a great product, by trying to be cheap at this point.

Le_Singe
:cool:

[GZsound]

That is an excellent answer.

I have been in your same position and even when using the same voice recorded at a different time, it can be difficult to recreate the EXACT sound of the previously recorded voice part. You need the same voice, the same mic, the same room, the same signal chain, etc. as the original recording and since our voices change from day to day, even having the same guy do the same part may not be a guarantee it will sound the same.

Since each voice is unique, it is nearly impossible to make one voice sound like another.. Which in a way is a good thing or every novice in his bedroom could put out vocal tracks that sound exactly like the professional vocalist on his favorite CD.

Maybe try and find the two words, or words that will work, in the already recorded narrative somewhere else and copy and paste them into the spot you need?

I don't think you would be happy with any attempt to make another voice sound like the voice you already recorded.

[robertruetz]

GZ has a good idea... is there ANYWHERE ELSE that the VO has the words you need. Some simple editing could do the trick if the two words are found elsewhere in the project. I'm constantly working with VO talent that either forgot to read something, or said something incorrectly. Sometimes, if I'm lucky, they've already said the words I need somewhere else--and I can just go a grab the piece and squeeze it where I need it.

Otherwise, your best bet for a match is still having the line re-cut, or having someone cheaper redo the entire VO for you.

For future reference, it's always a good idea to pay a little bit extra and build some revisions into your contract with any voice talent. You never know when you might need to revisit or correct something. It can at least save you some set-up charges.

Le_Singe
:cool:

[oretez]

depending on a whole lot of project specific variables . . . one of which is 'how close is close enough?'

with a little to a lot of work you can get close. And AA v2 includes frequency space editing which elevates it from 'duct tape' to 'bondo' for making repairs of this type

even if you find the words you need simply cutting and pasting will not produce seamless results (depending on 'how good is good enough?') . . .

not going to do full tutorial ('cause no two of these are exactly the same) . . . but don't ever work on original audio files . . . work on copies . . . work mutitrack (if possible)

one of the problems with just cutting and pasting is the audio bed underneath the 'words' . . . are you replacing dialog or adding material in the middle of something where there was no dialog . . . if replacing cut/paste is especially problematic

simply silencing the audio bed and punching or pasting in the new words will in most cases be just as obvious as using a new voice.

if it is possible that two voices will not noticeably detract from material, record a longer segment (from a point in the material where switching voices will not be as noticeable for example) and mix new material into the vid

rather then thinking about it as only 2 words think about feathering the segments together, sort of like spackle on a wall . . . then you can use some relatively subtle fx over an even larger segment (including material on either side of the splice) to help conceal the edit

depending on specific material all the tricks of the trade come into play, time fx, dynamic fx.

depending on the economics of situation you might a find VO guy willing to punch in a couple of sentences for about what it would cost you to buy software thus saving you your learning curve (and in my experience about the time you finish the first one of these you kind of learn what you should have done first (which is invariably different from what you did)

but, if you have more time then you know what to do with AA is as good as any tool for attempting this

[tech1]

I did similar things at a radio network I interned for several years ago. The host recorded a show in advance, and stumbled on a few words, but did not go back and repeat them, so there was an obviously really bad fumble in the middle of the introduction of the guest. I had to find matching words elsewhere, copy-n-paste them, and apply minute amounts of time-stretching on them so they fit in to the rhythm of the sentence. Then, i had to apply pitch-shifting to match the inflections of the rest of the sentence so it sounded natural. It was not 100% perfect, but sounded plenty good enough that no listener would question the integrity of the voice. The point is that it can be done, but you will need to use some tools to get it to sound natural. And I did this in AA 1.0 at the time.

[GZsound]

I actually don't understand why the process of cutting and pasting a narrative track is such a big deal..

I do voice over work all the time that requires editing and it is very simple to make a cut and pasted word sound exactly like the rest of the audio...after all, it IS exactly the same voice, just moved from one part of the track to another.

Many times the talent will stumble over a phone number, a name, an address, a city, pause too long between words, flub a word, etc. I simply have them repeat the part they flubbed and cut and paste it the way I want it to sound.

Sure, if you have a narrative that is imbedded in a backing track, the job gets very complicated quickly. But to simply cut and paste words in a single track of narrative shouldn't be intimidating at all.

[oretez]

the process of 'cut & paste' isn't (or shouldn't be) intimidating

whether it is appropriate choice depends on budget (time & $), source material (plus the editor's control over material), and what the specific goal might be

any or all of these variables will influence 'what's good enough'

and we (or I in this case) lack enough info (with regard to original poster) to make that determination

same room, same mic, same artist, same signal path on two successive days will produce noticeable (blind A/B) variations in same word (let alone phrase) but there are a number of instances when exercising digital magic is less satisfying (in terms of budget) then having someone punch in live (to a separate track of course) to correct an offending phrase . . .

One of the things that attracted me to digital editing was improved efficiency with this type of edit, but the same experience also suggests that, depending on the goal/target, that cut and paste is not always the tool for which to reach, particularly if you are forced to work with a mix and not source material. Particularly if you lack experience in digital domain and this is a 'one off' . . . an analog 'punch' can be far more intuitive, less complex to set up and less frustrating then attempting a crash course in digital

[GZsound]

I agree with all your points.

If you don't know what you are doing, don't do it.
If you have no experience in digital editing, don't do digital editing.
If you can hire the talent to do the part again, do so, but expect it to sound different than the original source material and plan on doing some digital manipulation to make it sound close.

If you don't know how to cut and paste, blend, pitch correct, stretch, etc. then don't do it.

If you have music imbedded in the file, it will be difficult if not impossible to cut and paste the narrative and have it fit.

So, the answer to his question is forget it. Try it. Don't try it. We aren't interested in giving you a tutorial on how to do it. Here are some suggestions on how to do it.

Gotta love internet forums.