VO-BB - 19 YEARS OLD!

[Dave S]

All, I had a problem recording my singing over a soundtrack. First of all, the provided soundtrack was at 80kbps, so I had to record my voice at that rate:
Question 1 - is that correct?
At first, things sounded OK. Then, after several playbacks, I heard this slightly metallic alien echo on my vocals. It seemed to get worse with time.
Question 2 - could this be the lossy 80 kbps rate?
The problem with that theory is that the soundtrack quality never degrades, just my vocals.

I've always just recorded my vocals, and mixed sound into that. This time I had a music soundtrack that I mixed into. I didn't know how else to do it, so I recorded my vocals in a separate file, then mixed into the soundtrack.
Question 3 - is that OK?

Question 4 - Could it be that all the tweaking I did on my voice file - EQ, Panning, normalizing, etc. have degraded the file due to the 80 kbps. I usually record at 256.

Sorry for the long post. Thanks in advance.

[Mushmelon]

It sounds to me as though you are working with MP3 files which are compressed files. If you repeatedly save an MP3 as an MP3 the end result will continue to degrade with each generation.
However, if you save the original MP3 as a wav (uncompressed) and work with wav files and continue to save as a wav file it won't degrade and won't sound any worse than the original did. Hope this makes sense.

Edit...I just re-read you post. You didn't have to record your vocals at the same bit rate as the soundtrack. If you save the sound track as an uncompressed wav file to start with you can record your vocals as an uncompressed wav as well and then mix them together in your audio application.

That being said, I am unsure why just your vocals would be degrading and not the soundtrack as well. Sorry
_________________
Tony Impieri
www.tonyimpieri.com

[sdelgo]

First off never ever ever normalize any mix. Normalizing makes your mix louder (which is what everybody wants) BUT it comes with the price of amplifying the noise floor as well. To get a nice loud mix... during tracking set your average level to -18db or so and record your tracks. Then mixdown to 2 tracks. Use multiband compression to tame any peaking freqencies or boost any lagging frequencies. Then to make the mix stand out use a limiter to bring the average level up to where you need it.
As for the 80kbps vocal track... by EQing the vocal track you're boosting the vocal frequencies and the inherent noise from the surrounding frequencies due to the fact that it's a low conversion rate... If your soundtrack is at 80kbps and you need to add vocals... record your vocals at 160kbps then mix the 2 files and save at 128kbps. I'm not sure if that will be what you're looking for but atleast you're not getting a bad overall sound due to file degredation. Hope this helps out.

Steve
_________________
you'll always have something on your plate... if you keep your bearings straight.

www.steviedproductions.com

[Jeffrey Kafer]

[sdelgo]

You can boost the lagging frequencies by the use of the make-up gain control.
A multiband compressor allows you to add compression (or not) to different groups of frequencies on your final mix. for example: you record a spot that has boomy lows but the mids and highs sound fine, you turn the overall level of the entire mix up in order to hear the high end but the lows peak and you really can't hear the mids or highs. You would use a multiband compressor to compress the low freqs and leave the mids and highs untouched allowing you to turn up the entire mix so the highs and mids are at the right level. So you're really not compressing the entire mix you're just compressing the part of the spectrum thats giving you problems. Don't mistake compression for normalizing. normalizing takes the highest level and boosts it to 0db and boosts everything else by the same amount. So if you peak at -9db and normalize... you'll boost everything by +9db not good if you're under the noise floor by -3db-you'll have +6db of added noise.
You're right about the limiter clipping off the high peaks BUT thats not the intended use of the limiter. You're overall level for your mixdown should be at or average out at -18db or so into your soundcard to avoid "digital artifacts" Digital equipment usually has a headroom of -18db with 0db being the clipping point. a limiter will allow you to set an absolute brickwall as to high how your signal will go. Once you have your mixdown at -18db or so and you start adding eq and the like you're boosting the level from the -18db on up... the limiter will not allow your signal to go past a pre determined level, thats when it will start "squashing" the dynamics of your mix. I hope i cleared some of this up for you. If you have anymore questions let me know. I really love this stuff. As you most likely can see from my site, I don't do too much in the VO character type stuff like alot on the board here. I do mostly radio imaging, where I create my own sound efx and beds and run my vocals through tons of processors in order to get "that voice" so I know a few things about getting stuff to sit in a mix.

Steve
_________________
you'll always have something on your plate... if you keep your bearings straight.

www.steviedproductions.com

[sdelgo]

Wow !!! Sorry for the gramatical errors on my previous post.

Steve
_________________
you'll always have something on your plate... if you keep your bearings straight.

www.steviedproductions.com

[sdelgo]

sheez gramatical = grammatical
_________________
you'll always have something on your plate... if you keep your bearings straight.

www.steviedproductions.com

[Jeffrey Kafer]

ah, ok, so you do boost the signal when you compress. My usual process goes like this:

1) Compress with 0 gain (I'll post my compression settings when I get home)
2) noise reduction
3) normalize to 0dB.

Is this not a good way to do it?
_________________
Jeff
http://JeffreyKafer.com
Voice-overload Web comic: http://voice-overload.com

[Dave S]

OK, I'm redfaced. Tony, you were right - they were mp3 files, and I should have saved then as wav and worked with them then. (What little I know, I knew that, where was my brain?) That would have helped.
Steve, I didn't normalize the mix, just the separate file before mixing. After that, you lost me. What do you mean by getting my mixdown to -18db. Do you mean my recording levels shouldn't go above -18?

Thanks for the input

Dave S

[sdelgo]

Dave,
When multi-tracking, the levels on your DAW should be at -18dbfs with some transient peaks a little higher no matter how many tracks you're going to use. Being at -18dbfs is pretty much "line level" or 0dbVU.
Example: 3 track VO session with music bed...
Track 1 - vocals -- record vocals at or near -18dbfs with eq, reverb etc... DO NOT normalize.
Track 2 - sfx -- record your synth sound efx at or near -18dbfs with eq, reverb etc...DO NOT normalize
Track 3 music bed-- record your music bed (usually pre-recorded) at or near -18dbfs DO NOT normalize

Now mix down all 3 tracks to 1 stereo track. Your level now should be anywhere
from -15dbfs to -9dbfs. Now is the time you would "sweeten" the mix using EQ,MultiBand Compression and or Limiter.
Few things to remember...
1. Never Normalize
2. dbfs stands for "decibels full scale" which is Digital
3. dbVU stands for "decibels volume units" which is Analog
4. -18dbfs = 0dbVU
5. MultiBand compressor can split up the sound spectrum and compress individual segments within.
6. Single Channel compressor can only compress the entire spectrum at once.
_________________
you'll always have something on your plate... if you keep your bearings straight.

www.steviedproductions.com

[sdelgo]

Jeff,

I'll wait for your compression settings...
But I have to ask... Why normalize the file if you compressed it earlier?

Steve
_________________
you'll always have something on your plate... if you keep your bearings straight.

www.steviedproductions.com

[Jeffrey Kafer]

[Jeffrey Kafer]

my compression settings which usually shave -3 to -6 dbs off the peaks


_________________
Jeff
http://JeffreyKafer.com
Voice-overload Web comic: http://voice-overload.com

[sdelgo]

Jeff,
i guess what I'm trying to get at is this... You're right about using compression to bring the peaks down to the lower signals. Then instead of normalizing you should use a limiter to bring the entire mix up. In limiting you can set a threshhold above the overall noise floor so you don't amplify the noise too. Normalizing will bring up the noise as well as the parts that you want to make louder. Normalizing will take the highest peak that is below 0dbVU and boost it right to 0dbVU but it will do the same for the rest of the mix. So if it needs to boost the highest peak by 3db to reach 0dbVU it will also boost everything (noise included) by the same.
Pretty much in any recording there is no need to normalize any signal.
But if it is working for you then thats great. I used to do it too until I was shown the results on a spectrum analyzer etc... by a mastering engineer. Heck, a majority of the stuff recorded by people on this board is for broadcast anyway so radio/tv stations compress and limit the heck out of the material anyway to make it fit the narrow bandwith of radio/tv. I guess I'm more into preserving the dynamic range as much as possible, as in music recording.
I'll see if I can locate that recording tutorial I was referring to and post it up here.
Steve
_________________
you'll always have something on your plate... if you keep your bearings straight.

www.steviedproductions.com

[jrodriguez315]

I must admit, for most of this thread, I have been completely lost, but it's fascinating nonetheless. I need to set up my studio and start playing with this stuff. Thanks so much for sharing your knowledge.

Major Procrastinator reporting for duty!
_________________
Joe Rodriguez, Bilingual Voice Actor | The Voiceover Thespian Blog