What Levels When First Opening Audio File in NLE? at DVinfo.net

Peter Moretti · February 19th, 2009, 02:54 AM

A friend used a SD 302 linked to a SD 744T to record dialogue as 24-bit .WAV files. I'm opening some of the audio files in Avid and am wondering if the recording levels were set properly.

I assume that the peaks should be around -4 on the digital scale. Is that about right?

The material is intended for web delivery.

Thanks much as always!

Petri Kaipiainen · February 19th, 2009, 04:41 AM

In the original file the dialog level should hover around -20 to -12 dBFS*. When you make the final file for net delivery use an audio compressor and raise the peak levels near 0 dBFs, maybe -3 dBFS peaks. That makes sound more intelligible on crappy PC speakers.

*) If it was recorded at 24 bits with SD302 & SD744 with good mic you can rise the levels almost 30 dB with no ill effects, if the levels are too low.

Peter Moretti · February 19th, 2009, 07:18 AM

Petri, which I assume means that when he recorded the files, the meter on the recorder should have been hovering between -20 and -12 dBFS, correct?

Petri Kaipiainen · February 19th, 2009, 07:23 AM

Yes, and that should be the level your editing systems shows. As digital recording systems are kind of locked to 0 dBFS, that should be the same and the levels are counted down from that no matter what the bit depth that was used when recording.

Can you analyze the dialog track WAV files with something which tells the real levels, like the statistical analyze tool in Audition? I have found that to be extremely helpfull and educating.

Peter Moretti · February 19th, 2009, 08:20 AM

I have Avid and Vegas, perhaps Vegas has such a feature.

I just want to be sure I understand what you mean by hovering between -20 and -12dBFS. That means the majority of the dialogue falls between -20 and -12dBFS. Not that the momentary peaks don't exceed -12dBFS, correct?

I ask b/c if the levels are set so peaks never exceed -12dBFS, it seems that a lot of dynamic range would be wasted.

Rick Reineke · February 19th, 2009, 09:19 AM

Audio levels for acquisition and final delivery are two different games. Obviously one must allow antiquate headroom for unexpected peaks in the field whilst back in post one can optimize via EQ, comp, limiting, ect. Field levels normally peaking around -12dBFS or so would be fine. (even in 16bit)

Vegas has good control and tools for optimizing the audio in post. fyi, Vegas was an audio NLE before it was video

Peter Moretti · February 19th, 2009, 09:39 AM

Rick, please let me expalin the source for my confusion:

While recording two minutes of dialogue, the peaks will usually stay at a pretty consistent level. But there will be one or two very brief moments when the dialogue gets considerably louder.

When you say to set levels so peaks are at -12dBFS, do you mean the peaks that the lion's share of the dialogue produces, or are you talking about the peaks related to the momentary bursts that can be quite a bit louder?

BTW, with the recording that my friend is doing, there are usually a few run throughs before recording. So he has a good idea what the "average peaks" will be and what the "momentay burst peaks" will be.

If, like in this case, it's reasonably known what the loudest level will be, why not just set the levels so the loudest level doesn't clip (and back off for a little leadway)?

Rick Reineke · February 19th, 2009, 02:26 PM

<i>"If, like in this case, it's reasonably known what the loudest level will be, why not just set the levels so the loudest level doesn't clip (and back off for a little leadway)?"</i>

Yes, if you know where the loudest peak will occur, set that very close to @ 0dBFS without going over. (clip) Or you could ride gain a little when the loud segments occur and/or use the mixers limiter.
Theoretically, in a perfect world, without using a limiter, 0dBFS would give the highest resolution, dynamic range and signal to noise.
My previous statement on levels was for general field production where unexpected loud peaks happen all the time without warning which requires additional headroom.

February 19th, 2009, 02:54 AM	#1
Peter Moretti Trustee Join Date: Mar 2007 Location: Sherman Oaks, CA Posts: 1,259	What Levels When First Opening Audio File in NLE? A friend used a SD 302 linked to a SD 744T to record dialogue as 24-bit .WAV files. I'm opening some of the audio files in Avid and am wondering if the recording levels were set properly. I assume that the peaks should be around -4 on the digital scale. Is that about right? The material is intended for web delivery. Thanks much as always! Last edited by Peter Moretti; February 19th, 2009 at 04:01 AM.

February 19th, 2009, 04:41 AM	#2
Petri Kaipiainen Major Player Join Date: Apr 2007 Location: Espoo Finland Posts: 380	In the original file the dialog level should hover around -20 to -12 dBFS. When you make the final file for net delivery use an audio compressor and raise the peak levels near 0 dBFs, maybe -3 dBFS peaks. That makes sound more intelligible on crappy PC speakers. ) If it was recorded at 24 bits with SD302 & SD744 with good mic you can rise the levels almost 30 dB with no ill effects, if the levels are too low.

February 19th, 2009, 07:18 AM	#3
Peter Moretti Trustee Join Date: Mar 2007 Location: Sherman Oaks, CA Posts: 1,259	Petri, which I assume means that when he recorded the files, the meter on the recorder should have been hovering between -20 and -12 dBFS, correct?

February 19th, 2009, 07:23 AM	#4
Petri Kaipiainen Major Player Join Date: Apr 2007 Location: Espoo Finland Posts: 380	Yes, and that should be the level your editing systems shows. As digital recording systems are kind of locked to 0 dBFS, that should be the same and the levels are counted down from that no matter what the bit depth that was used when recording. Can you analyze the dialog track WAV files with something which tells the real levels, like the statistical analyze tool in Audition? I have found that to be extremely helpfull and educating.

February 19th, 2009, 08:20 AM	#5
Peter Moretti Trustee Join Date: Mar 2007 Location: Sherman Oaks, CA Posts: 1,259	I have Avid and Vegas, perhaps Vegas has such a feature. I just want to be sure I understand what you mean by hovering between -20 and -12dBFS. That means the majority of the dialogue falls between -20 and -12dBFS. Not that the momentary peaks don't exceed -12dBFS, correct? I ask b/c if the levels are set so peaks never exceed -12dBFS, it seems that a lot of dynamic range would be wasted.

February 19th, 2009, 09:19 AM	#6
Rick Reineke Inner Circle Join Date: Aug 2008 Location: New York Posts: 2,039	Audio levels for acquisition and final delivery are two different games. Obviously one must allow antiquate headroom for unexpected peaks in the field whilst back in post one can optimize via EQ, comp, limiting, ect. Field levels normally peaking around -12dBFS or so would be fine. (even in 16bit) Vegas has good control and tools for optimizing the audio in post. fyi, Vegas was an audio NLE before it was video

February 19th, 2009, 09:39 AM	#7
Peter Moretti Trustee Join Date: Mar 2007 Location: Sherman Oaks, CA Posts: 1,259	Rick, please let me expalin the source for my confusion: While recording two minutes of dialogue, the peaks will usually stay at a pretty consistent level. But there will be one or two very brief moments when the dialogue gets considerably louder. When you say to set levels so peaks are at -12dBFS, do you mean the peaks that the lion's share of the dialogue produces, or are you talking about the peaks related to the momentary bursts that can be quite a bit louder? BTW, with the recording that my friend is doing, there are usually a few run throughs before recording. So he has a good idea what the "average peaks" will be and what the "momentay burst peaks" will be. If, like in this case, it's reasonably known what the loudest level will be, why not just set the levels so the loudest level doesn't clip (and back off for a little leadway)?

February 19th, 2009, 02:26 PM	#8
Rick Reineke Inner Circle Join Date: Aug 2008 Location: New York Posts: 2,039	<i>"If, like in this case, it's reasonably known what the loudest level will be, why not just set the levels so the loudest level doesn't clip (and back off for a little leadway)?"</i> Yes, if you know where the loudest peak will occur, set that very close to @ 0dBFS without going over. (clip) Or you could ride gain a little when the loud segments occur and/or use the mixers limiter. Theoretically, in a perfect world, without using a limiter, 0dBFS would give the highest resolution, dynamic range and signal to noise. My previous statement on levels was for general field production where unexpected loud peaks happen all the time without warning which requires additional headroom.