Distortion

Hi everyone,

I am back here with a question for audio. What is the best filter to adjust/fix/manipulate vocal sounds/voice that are distorted - (maxed out on volume)? Thank you.

John
FCP5, OSX10.4

none. once digital audio splats, it splats. thats more or less what youve got. all you can do (shy of really big bucks, and even then) is bring the splats down in volume

Hi John.

Whether there's anything that can be done depends on a number of factors, including how badly your audio is "clipped", how good you need your final audio to sound, and how much time and/or money you're willing to spend. If you have dialogue that's distorted at only a few places, you can often fix it by appling a low-pass filter to the segments that are distorted.

I've also saved several tracks by expanding out the audio waveform to show individual samples, and then actually "re-drawing" the waveform where the audio is clipped - this works extremely well, but is quite time consuming.

If you can arrange for us to listen to or download a short segment of your problem audio, we might be able to tell you more.

Travis
VoiceOver Guy and Entertainment Technology Enthusiast
[www.VOTalent.com]

As cliche as it may sound, I'm going with "it depends on how badly your audio is distorted."

I found out that running a clicks/pops fix on little distorted splats, can smooth out distortion slightly.

it's like blown out images, once they are gone they are gone, there is no info to grab.

It may be wrong but I would rather have audio that was recorded too low, i can at least bring the gain up with filters. That doesn't help you, but the next step is finding a protools guy i guess.

www.markdavid.tv

it's like blown out images, once they are gone they are gone, there is no info to grab.

It may be wrong but I would rather have audio that was recorded too low, i can at least bring the gain up with filters. That doesn't help you, but the next step is finding a protools guy i guess.

www.markdavid.tv

Agreed. All you can do is to help smoothen out the sound slightly.

Say, Travis, how do you actually work with the individual samples? it doesn't seem to make sense when i zoom in that deep. How do you recognize what you're doing?

Loud audio is preferable, but distortion is a big nono. Soft is better than distortion, but loud is better than soft.

Hi strypes,

If you put the file into the "file edit" window in Soundtrack Pro, and zoom in far enough you will see the individual samples, indicated by black squares on the waveform. You will see that any audio that is "clipped" because it is too loud will have sharp edges on the waveform at the point it reaches maximum level. This is where the audio has "hit the ceiling" - the waveform cannot extend any further because there aren't any more bits available to make the waveform louder.

If you grab the points in the sample and redraw the waveform to round off the edges, that harsh crackling sound will go away... Of course that means fixing each individual wave, which is why it's so time consuming.

Travis
VoiceOver Guy and Entertainment Technology Enthusiast
[www.VOTalent.com]

...and by the time you realize you need the rescue package, it's too late.

Yes, you can approximate an occasional peak repair here and there that "didn't make it," with the graphic tools but if it happens every other word--or every word, you have no show. You will be there for months fixing the waveforms, and you risk changing the character of the speaker if you do it too much.

I had a show that I was able to semi-rescue not by performing magic, but by noticing that the system was so overblown that the wrong microphone--the one on the non-speaking guest--actually sounded much better than the "right" one.

The other sound oddity was the widey circulated rumor of the programmer that had a program called "Clip Repair" or something like that. His web site was all but Slash-Dotted by the people trying to rescue their trashed audio tracks.

I contacted him and it turns out he had designed a way to more accurately recognize pops on a phonograph record and delete them. He admitted (with a red face visible through the email) that the program name might have been a little awkward.

Most of the major audio packages have a tool for "Clip Restoration," but nobody is beating down doors to use them more than once. I see several forums devoted to the academic exercise of trying to restore the top end of a waveform just by knowing the leading and lagging slopes--and how bad most approximations sound in the face of some very entertaining math.

This would all work much better if we could get people to talk in pure sine waves. You think we can get that?


You can get killed if the microphone amplifier clips and not the digital system. That will not give you a nice clean "clip" point to repair. That will give you saturation and harmonic garbage, not a flat top, and it's much harder to repair--or find.

Koz

Hi guys,

It seems like there's not much I could do, doesn't it? And I thought I didn't try enough.

It was a voice of a man I interviewed. His voice jumped quickly from low, deep to loud and high unpredictably , yet naturally. Lowering the volume helped. Low-pass helped. But the distortion is still there if I want to keep the integrity of the voice. Again, we are talking about perfection or close to perfect cleansing/fixing here.

You guys are great. Thank you all though. If I find something, I will post it up here.