Dialogue mixing at DVinfo.net

Ron Guilmette · May 17th, 2006, 11:20 AM

Anyone do this type of work?
I'm looking for some pointers on keeping the levels between the talent consistant.
I'm using some compression to keep the dynamics of each voice smooth without any real big volume swing.

What is the best approach to level matching from one person to the next? other than what I have been doing, listening over and over making fine adjustments.

This is a farely large project so they way I've been doing it takes more time than I have. Deadline fast approaching.........

Thanks!

Jimmy McKenzie · May 17th, 2006, 12:04 PM

The VST multiband compressor effect in Audition is great at performing this balancing act. Just knock down the obvious spikes that occur with words beginning with short "a" as in the word "at" so you have a level waveform prior to the compression.

Ron Guilmette · May 17th, 2006, 12:27 PM

I'm using the waves L1 to knock down the peaks to smooth out each voice on there respective tracks.
Are you adding the multi-band to the main buss to even out the volumes between talents?

Ron Guilmette · May 17th, 2006, 01:14 PM

The dialogue tracks are interviews of nine or so people. All with very different dynamics. A deep voice has more of a apparent volume than some others in the feature. Some spoke with more attitude than others.

Listening and adjusting volumes gets me close but not as close as I'd like to be.
Watching the peak meters doesn't help that much either. It would be much easier if I could see "average levels" insted.

Any ideas??

Seth Bloombaum · May 17th, 2006, 05:12 PM

Generally, you should be more concerned with perceived volume than with the meters when balancing loudness between different speakers.

Keeping or moving each speaker's audio to their own track can speed this along. Apply track level volume, eq and compression for each speaker based on the characteristics of their voice.

Then, use L1 on the master bus... When you think you have it, be sure to listen to it at a low playback level as well as medium and loud and see how it holds up.

That's the kind of workflow I'd use, my nle supports unlimited audio tracks and handy keyboard shortcuts to move a clip from track to track in perfect sync, your mileage may vary.

Marc Aitken · May 23rd, 2006, 06:41 PM

Hi Ron just came accross your post so you may be done now. But for future ref, I noticed you use the L1, so maybe you have a Waves bundle /package? if so see if you have the Paz analyzer or waves meters plug ins bundled. they have a great meter that I use all the time for voiceover and film ADR level matching. the beauty is that the two outside orange meterrs are peaking, but the single blue inner meter is average level. this combined with the peak hold readout has been invaluable to me as sometimes it helps to have a bit of "evidence" to back up that sneaky feeling that one was louder than the other!.

happy comping

regards
Marc

Steve House · May 24th, 2006, 04:07 AM

Quote:

Originally Posted by Marc Aitken

Hi Ron just came accross your post so you may be done now. But for future ref, I noticed you use the L1, so maybe you have a Waves bundle /package? if so see if you have the Paz analyzer or waves meters plug ins bundled. they have a great meter that I use all the time for voiceover and film ADR level matching. the beauty is that the two outside orange meterrs are peaking, but the single blue inner meter is average level. this combined with the peak hold readout has been invaluable to me as sometimes it helps to have a bit of "evidence" to back up that sneaky feeling that one was louder than the other!.

happy comping

regards
Marc

Which meters plugin are you referring to? The Paz Analyser or something else?

Mark Armstrong · May 24th, 2006, 05:07 AM

Ron

The only effective way to moniter dialogue is to have a mixer and another person monitering the sound ie a sound recordist.

Berns Ortiz · May 24th, 2006, 10:19 AM

I would suggest to judge loudness by ear, but if your ears are not tuned enough, get an "average" level meter (RMS, VU) plugin that you can use on your editor. Here's a free very good one, old school style:
http://www.pspaudioware.com/plugins/vmeter.html

First check to see if rumble and excesive or unwanted low freq exists. High pass filters set to 60-100 hz would be the tool. This is very crucial because excesive LF tends to trigger compressors off very easily when unneeded.
The trade off with compressors is that by getting more uniform levels, you are also sacrificing dynamic range, and hence the noise is going to go one way only, that's up. This is why its crucial to have good recorded dialog from the begining because the better it is from there, the more easy it is to tweak down the line, and the better the quality.
Also, its not a bad idea to use 2 compressors in series, or a compressor and limiter. The first compresor can catch only the peaks and bring them down a bit, so that the second compressor can take the duty of balancing in a better way. If using compresor instead of limiter, set the ratio of the peak compresor high (10:1 or more) and set the threshold so It ONLY catches the peaks!

It doesn't hurt to use a LP filter as well to cut hiss, and keep things neat.

the range from 3k-4k is were speech inteligibility is, so when you're recordings are laking in this aspect, a little bump around this part with a parametric EQ will make things sound better.

Jeff Riah · June 3rd, 2006, 12:39 PM

Has anyone used the PAZ Analyzer/Meters plugin to figure out a dialnorm setting for making DVDs?

The boss wishes me to do so but is it really as simple as soloing some DIA and running the PAZ meters on it in RMS with A-weighting and then use the BLUE meter?

Thanks.

May 17th, 2006, 11:20 AM	#1
Ron Guilmette Regular Crew Join Date: Jun 2004 Location: Hope Vally RI Posts: 30	Dialogue mixing Anyone do this type of work? I'm looking for some pointers on keeping the levels between the talent consistant. I'm using some compression to keep the dynamics of each voice smooth without any real big volume swing. What is the best approach to level matching from one person to the next? other than what I have been doing, listening over and over making fine adjustments. This is a farely large project so they way I've been doing it takes more time than I have. Deadline fast approaching......... Thanks!

May 23rd, 2006, 06:41 PM	#6
Marc Aitken New Boot Join Date: May 2006 Location: Bath UK Posts: 7	"seeing" average levels Hi Ron just came accross your post so you may be done now. But for future ref, I noticed you use the L1, so maybe you have a Waves bundle /package? if so see if you have the Paz analyzer or waves meters plug ins bundled. they have a great meter that I use all the time for voiceover and film ADR level matching. the beauty is that the two outside orange meterrs are peaking, but the single blue inner meter is average level. this combined with the peak hold readout has been invaluable to me as sometimes it helps to have a bit of "evidence" to back up that sneaky feeling that one was louder than the other!. happy comping regards Marc

May 24th, 2006, 05:07 AM	#8
Mark Armstrong New Boot Join Date: Mar 2006 Location: Sydney Australia Posts: 8	dialogue Mixing Ron The only effective way to moniter dialogue is to have a mixer and another person monitering the sound ie a sound recordist.

June 3rd, 2006, 12:39 PM	#10
Jeff Riah Tourist Join Date: Jun 2006 Location: Regina, SK Posts: 1	Using PAZ to calculate dialnorm? Has anyone used the PAZ Analyzer/Meters plugin to figure out a dialnorm setting for making DVDs? The boss wishes me to do so but is it really as simple as soloing some DIA and running the PAZ meters on it in RMS with A-weighting and then use the BLUE meter? Thanks.

May 17th, 2006, 12:04 PM	#2
Jimmy McKenzie Major Player Join Date: Jan 2004 Location: Waterloo Ontario Posts: 721	The VST multiband compressor effect in Audition is great at performing this balancing act. Just knock down the obvious spikes that occur with words beginning with short "a" as in the word "at" so you have a level waveform prior to the compression.

May 17th, 2006, 12:27 PM	#3
Ron Guilmette Regular Crew Join Date: Jun 2004 Location: Hope Vally RI Posts: 30	I'm using the waves L1 to knock down the peaks to smooth out each voice on there respective tracks. Are you adding the multi-band to the main buss to even out the volumes between talents?

May 17th, 2006, 01:14 PM	#4
Ron Guilmette Regular Crew Join Date: Jun 2004 Location: Hope Vally RI Posts: 30	The dialogue tracks are interviews of nine or so people. All with very different dynamics. A deep voice has more of a apparent volume than some others in the feature. Some spoke with more attitude than others. Listening and adjusting volumes gets me close but not as close as I'd like to be. Watching the peak meters doesn't help that much either. It would be much easier if I could see "average levels" insted. Any ideas??

May 17th, 2006, 05:12 PM	#5
Seth Bloombaum Inner Circle Join Date: Sep 2003 Location: Portland, Oregon Posts: 3,420	Generally, you should be more concerned with perceived volume than with the meters when balancing loudness between different speakers. Keeping or moving each speaker's audio to their own track can speed this along. Apply track level volume, eq and compression for each speaker based on the characteristics of their voice. Then, use L1 on the master bus... When you think you have it, be sure to listen to it at a low playback level as well as medium and loud and see how it holds up. That's the kind of workflow I'd use, my nle supports unlimited audio tracks and handy keyboard shortcuts to move a clip from track to track in perfect sync, your mileage may vary.

May 24th, 2006, 10:19 AM	#9
Berns Ortiz Regular Crew Join Date: May 2006 Location: Mexico Posts: 32	I would suggest to judge loudness by ear, but if your ears are not tuned enough, get an "average" level meter (RMS, VU) plugin that you can use on your editor. Here's a free very good one, old school style: http://www.pspaudioware.com/plugins/vmeter.html First check to see if rumble and excesive or unwanted low freq exists. High pass filters set to 60-100 hz would be the tool. This is very crucial because excesive LF tends to trigger compressors off very easily when unneeded. The trade off with compressors is that by getting more uniform levels, you are also sacrificing dynamic range, and hence the noise is going to go one way only, that's up. This is why its crucial to have good recorded dialog from the begining because the better it is from there, the more easy it is to tweak down the line, and the better the quality. Also, its not a bad idea to use 2 compressors in series, or a compressor and limiter. The first compresor can catch only the peaks and bring them down a bit, so that the second compressor can take the duty of balancing in a better way. If using compresor instead of limiter, set the ratio of the peak compresor high (10:1 or more) and set the threshold so It ONLY catches the peaks! It doesn't hurt to use a LP filter as well to cut hiss, and keep things neat. the range from 3k-4k is were speech inteligibility is, so when you're recordings are laking in this aspect, a little bump around this part with a parametric EQ will make things sound better.