I have some distortion in my sound. Is there anyway to fix it or make it better?

I use Mac so - what do you think?

Can you post a sample of the distortion?

Here is two small files

on with vocal and dist

and one speach with dist.

http://homepage.mac.com/arvidunsgaard/sound/FileSharing30.html

Sounds to me like headroom clipping, which, if true, cannot be fully "fixed"

There have been a couple of good semi fixes been posted here, but their authors names escape me - reckon they'll pop up before long, or have a good trawl through this forum, the relevant posts aren't that old.

Another good way to find them is to Google - try "sound clipping" etc as the search criteria.


CS

Sounds to me like headroom clipping, which, if true, cannot be fully "fixed"

There have been a couple of good semi fixes been posted here, but their authors names escape me - reckon they'll pop up before long, or have a good trawl through this forum, the relevant posts aren't that old.

Another good way to find them is to Google - try "sound clipping" etc as the search criteria.


CS

What does it mean with headroom clipping?

The strange thing is that it not get clipped in the camera.

is where the signal level exceeds the ability of a circuit to deal with it and results, literally, in the peaks of the sine waves being "clipped" off. If it happens into an A/D converter the result is pretty nasty.

If, as you say, the signal isn't clipped/ mangled "on camera" I'm a bit mystefied (adding hastily that this is getting way out of my comfort zone and would be better dealt with by one of DVinfo's REAL sound guru's).

Can't shed any light onto the NLE either as I'm a PC bod.


CS

It may or may not be actual clipping per se but it sounds like the signal is too hot for the input and is overloading on peaks - hard to say just where in the signal chain it's occuring without more information.. Tell us a little bit about how you recorded this and what equipment you used.

Sounds to me like the previous assessment is correct, its mild distortion from an input being overdriven, how do you know its not the camera being over driven? if you are looking at meters on the camera, that may not be an accurate way to show clipping as most meters on devices are what is called an RMS (average) level and not necessary PEAK metering. can you explain your signal chain and what equipment you used? the distortion seems mild and you can probably reduce the effect. but sound is like baking, once the ingredients are in there they are impossible to take out, all you can do is try to hide or de-emphasize.

Just to expand on my previous explanation of "headroom clipping" if, of course, you're interested.

OK, take a pure sine wave. Put it through an amplifier with an amplification factor of say, 10. At the O/P you get a sine wave 10 times the magnitude of the I/P.

All well and good. However, any amplifier has an O/P "ceiling" beyond which it physically cannot go. Lets say our amplifier has a ceiling of +/ - 10, determined by the voltage rails it's functioning from and other factors.

So, we increase the size of the I/P sine wave till the O/P reaches this ceiling. Then we go beyond it. The O/P cannot go beyond +/- 10 so that portion of the peak that "should" go past it is slammed into the ceiling for the duration of that peak.

We now have a sine wave with a flat top. It is no longer a sine wave but a partial square wave. Continue to increase the I/P, so more and more of the signal is slammed into this ceiling.

Take this to the point where the O/P IS actually a pretty good square wave.

We no longer have an analogue sine wave, but a digital square wave. Analogue circuits really, really, do not like digital signals, and will do everything in their power to take the edge off that square, thus becoming a very effective multi band filter.

Run this square wave through a spectrum analyser. Guess what? Our once pristine single frequency sine wave is now litterly thousands of different sine waves of various harmonics of the original frequency, stretching through the entire spectrum. A square wave is, in fact, nothing more than an infinite number of sine waves added together.

So, the moment you drive an amp past it's ceiling, that horrible raspy sound you hear is harmonics of the original being produced in their thousands, the number directly proportional to the amount of "clipping" going on.

It can be produced by "too hot" a mic into a pre - amp, the wick being turned up too high on that pre amp and driving a subsequent stage into clipping etc etc etc.

Feed a square wave (or part thereof) into an A/D converter (as in camera audio to digital tape "a la" DV or HDV) and they get very huffy indeed.

The reason for the "headroom"? This is the margin you must allow in your overall signal level to allow for transient peaks in excess of the average signal level. The less headroom you allow (which usually gives the best signal to following stages) also leaves you less able to deal with peaks, which will, of course, collide with the ceiling, thus giving you more and more clipping.

Allow no headroom at all, and thus force every single frequency in your I/P signal to clip, you unleash a veritable shit storm of er, mess loose on following stages.


Just thought I'd share.


CS

Ps. Whew, got through that with not one mention of "Slew Rate". Must be getting better.

I use a Sennheiser MKH 50 into a Shure FP 24 and out to a PD170 Dv Camera.

I use the 0db peep to adjust the volume on the camera and then listen all the time on the camera sound.

From the Shure mixer it was line out to the PD170 camera, and line in on the PD170.

So what can I do?

Any plugins or something to get it better.

Thanks for your great help

If what we are hearing is, actually, clipping, there is not a whole heap to be done.

I am not convinced it could not have happened somewhere in post. The real DV audio guru's can help here, is there any way this is possible?

I never get this sort of problem, so am in total ignorance as to how it could come about, if, as you say, it's clean out of the camera.

If you are 100% certain this is most definately not on the "off camera" sound track, I'm stumped.


CS

Maybe I couldnt hear it - lucky me I also have a mic on the camera, so maybe I can mix it and use booth.

Im looking for a clipping restoration tool for osx - someone knowing?

I find some for windows not for osx.

If what we are hearing is, actually, clipping, there is not a whole heap to be done.
CS

Do you mean its impossible to fix?

Or do it better?

If it is, indeed, clipping, it is impossible to "fix".

It is possible to ameliorate (make better) but not possible to replace that which has been lost.

As I said, someone is going to home in here and give you that magic bullet to "ameliorate" it, but fix totally, no, it is simply not possible.

Figure out where it's happening and nail it, once and for all, before you shoot any more stuff that is important. Once the info's gone, it's gone.

CS

The strange thing is that it not get clipped in the camera.

Hi Arvid,

First the good news - I don't think your problem is clipping. I looked at the waveform on the first clip (church singing) and the signal looks fine. If you had a clipping problem, the wavetops would not be smooth sinusoidal tops but instead would look like flat-top mountains. You see, clipping distortion flattens audio waves - once the analog to digital conversion reaches peak, everything above the max is simply all digital ones. But your waveform doesn't have flat-tops, so you don't have a clipping problem. Here's a simple explanation of the A-D quantizing problem. Let's say I have 5 binary digits to guess anyone's age: the digits are 16-8-4-2-1. Your age is 29, in our base 10 that's 16+8+4+x+1, or in digital it's 11101. Now let's do your son who is 7 years old: he's 0+0+4+2+1, or digitally 00111. So far, so good. Now let's try your 52 year old dad - oops, the highest we can go is 16+8+4+2+1 (31 years old), or digitally 11111. And how about your grandfather? Again, 31 is tops, or 11111. You see the problem? Once our A-D converter makes all ones (flat-top square wave), anything higher is still all ones. Clipping (maxing or pegging the meter in red) is not desirable. For those in the know, we call this quantization error (running out of binary digits).

Anway, perhaps part of your problem was that your microphone was too far away from the talent. In that case, perhaps you turned up the audio gain to compensate for the problem. Unfortunately, you're picking up a lot of reflected waves bouncing off floors, walls, etc before reaching your microphone. At any rate, like the clipping problem your options are limited. But perhaps you might try filtering out some of the high frequency audio (low pass filter) which might sound a bit more pleasing.

My second concern was running a mixer output (hot line level signal, 50K ohm high impedence) directly to your PD170 (requires weak microphone level signal, 600 ohm low impedence). Perhaps you shut down the PD170 amplifier gain to try to compensate for the signal mismatch, but I'm guessing you've got an impedence mismatch problem.

Next time you may want to try something like the Matchbox DB25:
http://media.zzounds.com/media/db25man-5ee35f985af9b7c22ddf7b082f6c8864.pdf

Finally, if the audio sounded better in the camera, I'd guess the problem just isn't as perceptible - the camera speaker isn't a good test. If you aren't doing so already, be sure to wear a good set of headphones while monitoring the audio.

Good luck, Michael

Hi! Michael!

What can I say, you are awesome. I learn everytime im here.

Thanks.


Arvid

I use the 0db peep to adjust the volume on the camera and then listen all the time on the camera sound.

Arvid, when you use the 0 dB 1 Khz tone from your Shure FP24 mixer, are you adjusting the sound level of the camera to -12 dB ?

Digital recording devices ( including cameras ) have no "head-room" past 0 dB ( any sound louder than 0 dB on your camera may cause clipping distortion ), so normally you would record the 0 dB tone from your mixer at -12 dB on your camera so that you will have an extra 12 dB of "head-room" when you record.

Also, are you using high-quality headphones to monitor the sound you are recording with your camera?

Sony 7506 headphones are pretty much the standard for monitoring sound for video.

Thanks for all your help.

Ok, so I have some problem with my sound, not all the time but...

Yesterday I download waves that have something called clip restoration.

But what do you think i can do to make it better.

I found this:

http://www.mindspring.com/~mrichter/formula/formula.htm

Which tool step by step can I use to get it better, or sounds better?

Arvid, when you use the 0 dB 1 Khz tone from your Shure FP24 mixer, are you adjusting the sound level of the camera to -12 dB ?

Digital recording devices ( including cameras ) have no "head-room" past 0 dB ( any sound louder than 0 dB on your camera may cause clipping distortion ), so normally you would record the 0 dB tone from your mixer at -12 dB on your camera so that you will have an extra 12 dB of "head-room" when you record.

Also, are you using high-quality headphones to monitor the sound you are recording with your camera?

Sony 7506 headphones are pretty much the standard for monitoring sound for video.

I use a headphone that came with the mixer - I dont know the name.