Audio solution needed please:

I recorded a presenter that alternately shouted when he got excited or to emphasize his points, talked softly or moderated his speech level now and then. The direct result was a audio track that’s uneven to say the least.

I need to be able to smooth the whole thing out as boosting the individual sections that require it would be tedious and very time consuming. Adding gain universally to the whole track would of course result in even more distortion and spikes!

Considering the above, what is my best course of corrective action to make the best of the situation and to have something that would be broadcastable on cable? I don’t know all that much about audio. For tools I have Adobe Premiere 6.5, Pro 2 and Pro 3.
What applications, processes or combination of processes/procedures should be utilized? Free or low-cost downloads preferred.

It would be greatly appreciated if someone who has experienced a similar problem which must be fairly common I would think would explain to me in detail exactly how they addressed the problem to successfully resolve it.

Thanks for any practical advice or pointer offered.

Bruce

There really isn't any automated solution possible. You hit it on the head - you need to go through the production segment-by-segment, shot-by-shot, adjusting gain and equalization, reverb, etc, until it sounds exactly like you want it to sound. Yes, it will be time-consuming and nit-picky but that's why they pay sound mixers big bucks. You'll need to calibrate your workstation, use proper studio monitors, and work your way through the production balancing the various audio elements by ear. You can use Premiere, your mouse, and the track "rubber bands." A tool like a Mackie Control or similar interface that emulates a mixing desk can be a blessing, as will be your track automation tools, so that as you move the faders the system will remember the settings frame by frame and can replay them during the final rendering of the tracks.

...It would be greatly appreciated if someone who has experienced a similar problem which must be fairly common I would think would explain to me in detail exactly how they addressed the problem to successfully resolve it....

I use WaveLab as the sweetening tool...

My standard process would be to:
1) Normalize the track
2) remove DC offset
3) mild EQ adjustments after looking at Spectrum Analyzer (FFT), "Q" Pluggin (parametric equalizer)
4) Add Compressor plugin (3 to 1 ratio) -- "Clean Compressor" or "Multiband Compressor"...boost output gain to EXPAND the quieter stuff.

Adobe Soundbooth has a great filter that I think is under advanced compressors called "voice leveler". As long as nothing is clipped this tool works like a champ for me.

I tried Wave Lab for example and experienced the following;
1) A Wavelab 6 demo/trial does not appear to exist and the app is pretty costly.
1b) The Wavelab 5 demo is crippled and its usuablility fairly limited, can't save etc.

Also: The Adobe Soundbooth demo AFAIK only lasts for 2 days!

As a starting point I would appreciate knowing how to do the following with Adobe Premiere Pro 2.

If there are multiple ways of selecting in and out points on a timeline in regards to applying effects that's vital as well. Thanks for your patience and I apologize for my ignorance as this side of editing is entirely new to me.

Please, would either of you or anyone else be willing to explain in detail or list step by step how to:

1) Select the desired range on the timeline and to then
2) Select the effect to be applied only towards that range and to apply it successfully.

Thanks so much for your help.

Bruce

I'd be inclined to use SoundForge and their "normalize" tool.
With this tool, I select normalize to an RMS value with compression to -0.1 dBFS. You will, however, need to break up the track into segments representing fairly constant audio levels.

The Conversations Network has an amazing tool that was really designed for podcasters to do just that, called the Levelator (http://www.conversationsnetwork.org/levelator/). It works under Windows, Mac OS X, or Linux.

From my understanding, it analyzes the waveform of the audio, and then roughly averages it all out. It still keeps minor changes between soft and loud, so it still sounds realistic. It has really done an amazing job for me in the past! Although it can get a little confused when there's clapping (I only experienced that once though, recording for a school jazz band when people were clapping..I think in general this tool really shouldn't be used with music just because it narrows the "contrast" of the music's dynamics)

From my experiences, this tool has worked wonders! There was another time when I was working on a short video for my high school's television network, and the student field correspondent never moved the microphone away from her mouth when she was interviewing someone. So of course all of her audio came out just fine, but you could barely hear the interviewee. I ran the audio through the levelator and bam! It was perfect!

So, to use the program, all you have to do is export your audio to individual files, and then just drag and drop it onto the levelator, and it will do it's magic, then save your file with .output appended to it.

Best of all, it's free! That website again is: http://www.conversationsnetwork.org/levelator/

Best of luck!

I would use a well set up audio compressor!

Just a note of caution in all this talk of normalization and compression. Remember normalization raises everything to a uniform level. But dynamic range, variations in loudness, are what give life and substance to a performance. Remove all variability and make everything uniformly loud and you run the risk of a production that is dull and without any dramatic contrasts. Certainly the audience shouldn't have to strain to hear quieter passages or grabbing for the volume remote during loud ones but keep the normalization and compression within reason. Try to avoid the temptation to join pop music's loudness wars.

The Conversations Network has an amazing tool that was really designed for podcasters to do just that, called the Levelator (The Levelator from The Conversations Network (http://www.conversationsnetwork.org/levelator/)). It works under Windows, Mac OS X, or Linux.

From my understanding, it analyzes the waveform of the audio, and then roughly averages it all out. It still keeps minor changes between soft and loud, so it still sounds realistic. It has really done an amazing job for me in the past! Although it can get a little confused when there's clapping (I only experienced that once though, recording for a school jazz band when people were clapping..I think in general this tool really shouldn't be used with music just because it narrows the "contrast" of the music's dynamics)

From my experiences, this tool has worked wonders! There was another time when I was working on a short video for my high school's television network, and the student field correspondent never moved the microphone away from her mouth when she was interviewing someone. So of course all of her audio came out just fine, but you could barely hear the interviewee. I ran the audio through the levelator and bam! It was perfect!

So, to use the program, all you have to do is export your audio to individual files, and then just drag and drop it onto the levelator, and it will do it's magic, then save your file with .output appended to it.

Best of all, it's free! That website again is: The Levelator from The Conversations Network (http://www.conversationsnetwork.org/levelator/)

Best of luck!

Has anyone else tried using the levelator for video? I'm having a similar problem. I have a 22 minute thing, and I can't seem to get the audio levels just right. And I don't want to take it to a sound guy if I don't have to. ...I was thinking of just saving the audio for the entire thing as an mp3, then sticking it in the levelator and seeing what happens. Does anyone else know of any other tools like this? Things that automatically adjust the levels?

Has anyone else tried using the levelator for video? I'm having a similar problem. I have a 22 minute thing, and I can't seem to get the audio levels just right. And I don't want to take it to a sound guy if I don't have to. ...I was thinking of just saving the audio for the entire thing as an mp3, then sticking it in the levelator and seeing what happens. Does anyone else know of any other tools like this? Things that automatically adjust the levels?

I haven't tried it yet, but I will this weekend.

However, it would make sense just to try it. It's free. It couldn't be simpler to use (drop you audio file on it). Then listen to the results.

The only real test for audio is to listen to it, so try the Levelator, and if the audio sounds good, use it, if not, don't.

If you were in Final Cut, this would be as easy as opening the mixer and turning on keyframe automation. You listen, adjust levels as the clip goes by and you are done. So simple, so easy, so few video editors who use FCP even have a clue that they have a pretty decent little simple mixer built in.

Dan

you know we see this all the time. something is going to be hard time consuming work, is there a filter that will fix it ?

often the answer is no, you have to do the hard work. in this particular case, there *might* be a filter or two that can help, but doing what needs to be done ins't the end of the world, and its what seperates the pro's from the amatuers

normalizing won't solve anything, and is not something to just do. its basically a gain adjustment that brings everything up including noise. its something to use on the quite sections to bring them up to match the high sections, but just the normal level mixing will do the same, plus that is still changable.

you may find that just some compression will help. start at a lower threshold, say -30 or so, and use a 1.5:1 or 2:1 ratio and see how that works. then if its better, consider applying a second compressor with a higher threshold at -20 or -15, and a slightly higher compression ratio to bring the peaks down more gracefully - say 2:1 or even 3:1. that might do the job and bring the mix into a reasonable middle ground.

For Sony Vegas users, this is an easy fix using the audio plugin chain. Lots of options in there. Start out using normalize and the compressor.

I use Diamond Cut from Audio Restoration, Audio Forensics, and Noise Reduction Software. Audio Editing, DC FIVE. (http://www.enhancedaudio.com) -- it does a super job on things like this, and can also do many other things to improve the clarity and quality of your audio. Its intended application is the cleaning up of old audio (from 78s and vinyl LPs, for example), and it will make amazing audio out of some real crud. It can even repair clipped audio.

It can also stretch/shrink your audio to achieve sync at both ends of a clip if your video and audio didn't maintain sync over a long shot -- and it does it without changing pitch!

And LOTS of other neat stuff.

-- Carl

Somehow that URL got changed. It's www enhancedaudio com if I leave out the periods. Originally I typed it with the periods, and from my PC to the Board it got altered. The current version is DC 7, which I have not tried, but I have DC 6 and love it.