Normalizing Sound at DVinfo.net

Tim Bickford · January 7th, 2008, 08:14 AM

I recently discovered the "Normalize Sound" option in P-Pro-CS3. I'm not sure if this option was available on previous versions. I was hoping to get some help on using it properly.

I'm working on a project that utilizes dozens of different video clips shot from multiple cameras at different sites. The audio volume levels changes from clip to clip in my timeline. I've tried the "Normalize option" for each clip. Some clips are changed by as much as 9 db, while others by -1.5 db.

Q. What is the best way to normalize (or should I say) produce a video like this where there will be consistency in volume?

Thanks

Marty Baggen · January 7th, 2008, 10:49 AM

Normalizing is not a good method for consistent volume unless there is little dynamic range. Normalizing takes a waveform and raises it proportionally to peak level (or what ever level you choose, whether 0db, -1db, etc).

As an example.. let's say you have a 5 minute soundtrack, and it has one little peak that hits 0db. Normalizing will do nothing to your file. It sees the peak, and adjusts upward, dragging the whole waveform. If the peak maxed out at say, -1.5db, then the entire waveform is raised that amount.

For successful audio soundtrack work, you need to address the dynamics of your audio... limit your peaks, boost your low levels.

I do my sound work in a combination of Vegas 8 and Sound Forge. The supplied plug-ins aren't bad. Experiment with the Graphic Dynamic Range. Allow a max output of 0db, set your compression to around 3:1, and then adjust the threshold to your sound file.

Audio work is an oft forgotten element in our work.... many video producers would be surprised that a set of mastering audio plug-ins can cost more than their post production system and software.

Premiere is dysfunctional for serious audio work... not tried Audition.

Tim Bickford · January 7th, 2008, 12:46 PM

Marty-

Thank you for the info. I've focused so long on the video aspects of P-Pro that I've neglected to learn and understand the audio capabilities.

It appears as though I should strive to keep my audio peaks at 0 db? Although I'm still not sure. Also - is there a way with p-pro to scrub the audio and view the db level as it moves up and down?

When I get home I'll take a second look as your suggestions and try it out.

Thanks for the help.

Marty Baggen · January 7th, 2008, 01:11 PM

Quote:

Originally Posted by Tim Bickford

- is there a way with p-pro to scrub the audio and view the db level as it moves up and down?

When I get home I'll take a second look as your suggestions and try it out.

Thanks for the help.

Not in a fashion that would do you any good. You need to render out an audio file that you can look at in Sound Forge.

Try to get your hands on a professionally produced program of similar style to yours, and somehow get the unaltered version of the soundtrack into Sound Forge and just look at it. It will serve as an interesting guide.

Bart Walczak · January 8th, 2008, 03:32 AM

For simple audio work premiere track effects might suffice. You can add equalizer and compressor/limiter to your tracks and work with these. But for professional audio workflow you'd need something like Audition or Sound Forge.

Paul R Johnson · January 8th, 2008, 05:18 AM

Normalise is useful for getting those tracks recorded overall too low back up to something closer to the others. Depending on the content, some compression can be handy, or to deal with the odd really loud peak, a compressor and a limiter. You have to be carefull - background sounds leap up very strangly when you start playing with compressors - in an external audio hardware or software system you can also get things like a voice over automatically reduce the other audio to give it some space. The amount, time taken to drop and restore can all be tweaked.

I have to admit that in most cases, I stay within premiere and just rubber band the audio tracks and tweak till it sounds right.

Tim Bickford · January 8th, 2008, 10:59 AM

Marty-

I installed the trial version of sound forge. Would it be possible for you to write out some simple step by step instructions on how to make volume adjusments to an audio file in sound forge?

I plan to export a single audio file (from P-Pro) that is made up of 25 or 30 diffent audio clips combined. I.E. 1-2-3-4-5-6....29-30 = 1.

Any further assitance would be greatly appreciated.

Glenn Davidson · January 8th, 2008, 01:06 PM

Here is my technique for leveling out multiple regions of audio. I use Protools. I first go in to each segment and manually highlight and gain reduce major spikes. Then I normalize, not for peak but for RMS. This gets things pretty close. I then put markers on each segment and rapidly play them back in sequence at low volume to reveal their relative level. I will then apply gain, either plus or minus, to any segments that are too high or too low. The process can take days on some projects, but the results are worth it.

Marty Baggen · January 8th, 2008, 01:23 PM

Tim... you'll appreciate Sound Forge, and I assume Audition offers similar function, but I am not experienced with it.

My steps will vary slightly with your setup in that I use some 3rd party plug-ins, but the basic approach is the same.

1 - Once you have a lock on your edit, output each track (or stem) as a soundfile. You should keep your syncsound on a track, music on another, etc

2 - bring each stem into Sound Forge. Depending on the characteristics of the waveform, make sure you've got at least 3 or 4db of headroom from the highest peak from 0db. Lower or raise the overall volume accordingly. There may be instances where you do some dynamics work on the file if there are only a few high peaks.

3 - Now look at your waveform. Where are the majority of peaks? If you zoom out on the view, it will begin to form a more solid bar. Apply the Graphic Dynamics plug-in and set your threshold to encompass where that solid bar peaks. Maybe it's around -8 to -12db. Set your output to -2 or -3db, then set your compression to 3:1.

4 - Apply the Dynamics effect.

5 - Look at your waveform... does it still have nice sharp peaks, but less variance between the loud peaks and lower levels? Or, does the waveform look mashed and flattened? A flattened waveform is a combination of too low of threshold and/or too severe compression ratio.

6 - If you have professionally produced soundtrack music for your project, there's no need to mess with its dynamics.

7 - (Another variance in my workflow is that I do all my track mixing in Vegas 8, but if you have steady hands... you can achieve decent results in PPro). Mix your levels between tracks in the NLE... being careful not to clip on the peaks.

8 - Output your mix as a single waveform, then repeat the process above for the final dynamics and level... this time, setting the output level to 0db.

That's my workflow as adapted to the tools you have. Premiere is a beast to do any sort of mixing, but if you get the hang of it, it can be functional.

Trial and error is the key... and look at professional produced waveforms, maybe from a CD of a soundtrack or something that mirrors your own projects. You can learn a lot from just looking.

My methods are certainly not gospel.... so much depends on the tools you have at your disposal and what your objectives are... but good sound should always be a goal of video producers.

Tim Bickford · January 8th, 2008, 03:09 PM

Marty-

Thank you for taking the time to write all these steps out. I'll give this a try today and will let you know how I do.

Unfortunately my project has a deadline (very soon). I hope to be able to get through the basics and produce reasonable audio.

Thanks again!

Marty Baggen · January 8th, 2008, 03:27 PM

Tim... PM me if you run into any troubles.

You could even send a sample chunk of waveform, and I could document the settings I would use.

Tim Bickford · January 9th, 2008, 12:56 PM

Marty-

I did not see your PM offer until now. I tried your suggestions. It took me a while to get use to the software. Well... I should say get use to it enough to do the things that you suggested.

For the most part I was able to produce a much better sound file. I'm very happy with the results. I'll try to post the audio file in a place that you can download it. Perhaps you may have some settings that would make it better.

I'll get back to you.

Thanks again!

Marty Baggen · January 9th, 2008, 03:01 PM

Fantastic.... that's very great news. Keep it up Tim, and I look forward to hearing what you have come up with.

Ryan Postel · February 27th, 2008, 11:50 AM

Tim,

I realize is probably past your deadline, but if you did get the Production Suite for CS3, Adobe has easily integrated a workflow between Premiere and Audition. I wouldn't recommend doing any sound editing in Premiere. Sound Forge is a good alternative, as I'm sure you have found, but Audition has great integration to the point where if you tweak things in Audition, it will do a live update to the link in Premiere. Its a nice feature. But most audio editing programs are almost exactly the same now.

Jim Day · February 27th, 2008, 02:25 PM

There is a lot of good information in this thread (and I learned a lot). I'm not an audio expert by any stretch of the imagination but I do want to reiterate one point that was made above. One poster mentioned that normalize is good for clips that are (overall) recorded too low.

In addition to the video work I do, I also manage our church's website. We record the sermon each Sunday and I put it on the website for those who cannot make it on Sunday can listen to it. For some reason, the CD recorder we use (hooked up through our sound board) records the volume so low that it cannot even be heard. This has nothing to do with the settings on the recorder because they are maxed out. Anyway, to make these sermons usable, I simply import them into SoundForge and normalize them. It's a 1-step process that works great in this case.

January 7th, 2008, 08:14 AM	#1
Tim Bickford Major Player Join Date: Sep 2005 Location: New Hampshire Posts: 448	Normalizing Sound I recently discovered the "Normalize Sound" option in P-Pro-CS3. I'm not sure if this option was available on previous versions. I was hoping to get some help on using it properly. I'm working on a project that utilizes dozens of different video clips shot from multiple cameras at different sites. The audio volume levels changes from clip to clip in my timeline. I've tried the "Normalize option" for each clip. Some clips are changed by as much as 9 db, while others by -1.5 db. Q. What is the best way to normalize (or should I say) produce a video like this where there will be consistency in volume? Thanks __________________ Tim Bickford

January 7th, 2008, 12:46 PM	#3
Tim Bickford Major Player Join Date: Sep 2005 Location: New Hampshire Posts: 448	Marty- Thank you for the info. I've focused so long on the video aspects of P-Pro that I've neglected to learn and understand the audio capabilities. It appears as though I should strive to keep my audio peaks at 0 db? Although I'm still not sure. Also - is there a way with p-pro to scrub the audio and view the db level as it moves up and down? When I get home I'll take a second look as your suggestions and try it out. Thanks for the help. __________________ Tim Bickford

January 8th, 2008, 10:59 AM	#7
Tim Bickford Major Player Join Date: Sep 2005 Location: New Hampshire Posts: 448	Marty- I installed the trial version of sound forge. Would it be possible for you to write out some simple step by step instructions on how to make volume adjusments to an audio file in sound forge? I plan to export a single audio file (from P-Pro) that is made up of 25 or 30 diffent audio clips combined. I.E. 1-2-3-4-5-6....29-30 = 1. Any further assitance would be greatly appreciated. __________________ Tim Bickford

January 8th, 2008, 03:09 PM	#10
Tim Bickford Major Player Join Date: Sep 2005 Location: New Hampshire Posts: 448	Marty- Thank you for taking the time to write all these steps out. I'll give this a try today and will let you know how I do. Unfortunately my project has a deadline (very soon). I hope to be able to get through the basics and produce reasonable audio. Thanks again! __________________ Tim Bickford

January 9th, 2008, 12:56 PM	#12
Tim Bickford Major Player Join Date: Sep 2005 Location: New Hampshire Posts: 448	Marty- I did not see your PM offer until now. I tried your suggestions. It took me a while to get use to the software. Well... I should say get use to it enough to do the things that you suggested. For the most part I was able to produce a much better sound file. I'm very happy with the results. I'll try to post the audio file in a place that you can download it. Perhaps you may have some settings that would make it better. I'll get back to you. Thanks again! __________________ Tim Bickford

January 7th, 2008, 10:49 AM	#2
Marty Baggen Major Player Join Date: May 2005 Location: monroe, or Posts: 572	Normalizing is not a good method for consistent volume unless there is little dynamic range. Normalizing takes a waveform and raises it proportionally to peak level (or what ever level you choose, whether 0db, -1db, etc). As an example.. let's say you have a 5 minute soundtrack, and it has one little peak that hits 0db. Normalizing will do nothing to your file. It sees the peak, and adjusts upward, dragging the whole waveform. If the peak maxed out at say, -1.5db, then the entire waveform is raised that amount. For successful audio soundtrack work, you need to address the dynamics of your audio... limit your peaks, boost your low levels. I do my sound work in a combination of Vegas 8 and Sound Forge. The supplied plug-ins aren't bad. Experiment with the Graphic Dynamic Range. Allow a max output of 0db, set your compression to around 3:1, and then adjust the threshold to your sound file. Audio work is an oft forgotten element in our work.... many video producers would be surprised that a set of mastering audio plug-ins can cost more than their post production system and software. Premiere is dysfunctional for serious audio work... not tried Audition.

January 8th, 2008, 03:32 AM	#5
Bart Walczak Major Player Join Date: Apr 2006 Location: Warsaw/Poland Posts: 716	For simple audio work premiere track effects might suffice. You can add equalizer and compressor/limiter to your tracks and work with these. But for professional audio workflow you'd need something like Audition or Sound Forge.

January 8th, 2008, 05:18 AM	#6
Paul R Johnson Inner Circle Join Date: Nov 2005 Location: Lowestoft - UK Posts: 4,015	Normalise is useful for getting those tracks recorded overall too low back up to something closer to the others. Depending on the content, some compression can be handy, or to deal with the odd really loud peak, a compressor and a limiter. You have to be carefull - background sounds leap up very strangly when you start playing with compressors - in an external audio hardware or software system you can also get things like a voice over automatically reduce the other audio to give it some space. The amount, time taken to drop and restore can all be tweaked. I have to admit that in most cases, I stay within premiere and just rubber band the audio tracks and tweak till it sounds right.

January 8th, 2008, 01:06 PM	#8
Glenn Davidson Major Player Join Date: Dec 2005 Location: Nevada City, California Posts: 499	Here is my technique for leveling out multiple regions of audio. I use Protools. I first go in to each segment and manually highlight and gain reduce major spikes. Then I normalize, not for peak but for RMS. This gets things pretty close. I then put markers on each segment and rapidly play them back in sequence at low volume to reveal their relative level. I will then apply gain, either plus or minus, to any segments that are too high or too low. The process can take days on some projects, but the results are worth it.

January 8th, 2008, 01:23 PM	#9
Marty Baggen Major Player Join Date: May 2005 Location: monroe, or Posts: 572	Tim... you'll appreciate Sound Forge, and I assume Audition offers similar function, but I am not experienced with it. My steps will vary slightly with your setup in that I use some 3rd party plug-ins, but the basic approach is the same. 1 - Once you have a lock on your edit, output each track (or stem) as a soundfile. You should keep your syncsound on a track, music on another, etc 2 - bring each stem into Sound Forge. Depending on the characteristics of the waveform, make sure you've got at least 3 or 4db of headroom from the highest peak from 0db. Lower or raise the overall volume accordingly. There may be instances where you do some dynamics work on the file if there are only a few high peaks. 3 - Now look at your waveform. Where are the majority of peaks? If you zoom out on the view, it will begin to form a more solid bar. Apply the Graphic Dynamics plug-in and set your threshold to encompass where that solid bar peaks. Maybe it's around -8 to -12db. Set your output to -2 or -3db, then set your compression to 3:1. 4 - Apply the Dynamics effect. 5 - Look at your waveform... does it still have nice sharp peaks, but less variance between the loud peaks and lower levels? Or, does the waveform look mashed and flattened? A flattened waveform is a combination of too low of threshold and/or too severe compression ratio. 6 - If you have professionally produced soundtrack music for your project, there's no need to mess with its dynamics. 7 - (Another variance in my workflow is that I do all my track mixing in Vegas 8, but if you have steady hands... you can achieve decent results in PPro). Mix your levels between tracks in the NLE... being careful not to clip on the peaks. 8 - Output your mix as a single waveform, then repeat the process above for the final dynamics and level... this time, setting the output level to 0db. That's my workflow as adapted to the tools you have. Premiere is a beast to do any sort of mixing, but if you get the hang of it, it can be functional. Trial and error is the key... and look at professional produced waveforms, maybe from a CD of a soundtrack or something that mirrors your own projects. You can learn a lot from just looking. My methods are certainly not gospel.... so much depends on the tools you have at your disposal and what your objectives are... but good sound should always be a goal of video producers.

January 8th, 2008, 03:27 PM	#11
Marty Baggen Major Player Join Date: May 2005 Location: monroe, or Posts: 572	Tim... PM me if you run into any troubles. You could even send a sample chunk of waveform, and I could document the settings I would use.

January 9th, 2008, 03:01 PM	#13
Marty Baggen Major Player Join Date: May 2005 Location: monroe, or Posts: 572	Fantastic.... that's very great news. Keep it up Tim, and I look forward to hearing what you have come up with.

February 27th, 2008, 11:50 AM	#14
Ryan Postel Regular Crew Join Date: Feb 2008 Location: Chicago, IL Posts: 93	Tim, I realize is probably past your deadline, but if you did get the Production Suite for CS3, Adobe has easily integrated a workflow between Premiere and Audition. I wouldn't recommend doing any sound editing in Premiere. Sound Forge is a good alternative, as I'm sure you have found, but Audition has great integration to the point where if you tweak things in Audition, it will do a live update to the link in Premiere. Its a nice feature. But most audio editing programs are almost exactly the same now.

February 27th, 2008, 02:25 PM	#15
Jim Day Regular Crew Join Date: Jul 2007 Location: El Paso Posts: 35	There is a lot of good information in this thread (and I learned a lot). I'm not an audio expert by any stretch of the imagination but I do want to reiterate one point that was made above. One poster mentioned that normalize is good for clips that are (overall) recorded too low. In addition to the video work I do, I also manage our church's website. We record the sermon each Sunday and I put it on the website for those who cannot make it on Sunday can listen to it. For some reason, the CD recorder we use (hooked up through our sound board) records the volume so low that it cannot even be heard. This has nothing to do with the settings on the recorder because they are maxed out. Anyway, to make these sermons usable, I simply import them into SoundForge and normalize them. It's a 1-step process that works great in this case.