Subject:Distortion problems abound

Posted by: JML

Date:5/9/2003 12:21:30 PM

I'm running into problems with distortion in files that are nowhere near zero db. For example, I have a track which consists of swelling strings and dense pads that is giving me the most trouble. I tried normalizing this file to 98%... then 97%... down to 75%. Although the file was well below 0 db, I was getting distortion in the peaks. The same problem occurs when using the Wave Hammer plug-in. Even with auto-gain compensate turned off, I was unable to compress this file very much without lots of distortion.

My understanding of compression and normalization might be flawed, but I thought a file which is staying well below 0 db will not distort. Other songs which are more beat-driven and contain less high-energy, sustained elements don't seem to be as problematic, though I've still had distortion problems in these as well.

Might this be soundcard related? I'm currently using a consumer-grade Yamaha ds-xg. These files are at 24-bit, and I'm wondering if the files are fine, but my soundcard is not able to deal with them.

Subject:RE: Distortion problems abound

Reply by: Geoff_Wood

Date:5/9/2003 5:33:46 PM

It sounds like the distortion is occurring way before SForge, maybe at your soundcard input. Do you have a '+4' level running into a '-10' soundcard (pro v. consumer) ?

I'm not sure how you soundcard signal flow goes, but try reducing the analogue signal level into the card until no distortion occurs on record. Check the shape of the peaks at various input levels.

Once you have a good level, try again, this time with the mixer applet record volume increasing the level into the SF record monitor meters. Hopefully you will then get an undistorted signal at good volume.

Normalisation should not change the waveform peak shapes.

geoff

Subject:RE: Distortion problems abound

Reply by: JML

Date:5/10/2003 2:18:31 PM

I guess I should have mentioned at the outset that these files came from Reason, so it is all internally generated; no issues with a/d conversion.

The peak shapes aren't changing or clipping or anything. In fact, the whole file is staying well below 0db, which is why this is maddening. The file looks fine, the waveform isn't clipping, yet I am hearing distortion on playback. This happens after normalization and during real-time auditioning of Wave Hammer processing.

Subject:RE: Distortion problems abound

Reply by: Geoff_Wood

Date:5/11/2003 1:00:57 AM

What do the waveforms look like fresh out of Reason ? Normalising doesn't flatten tops.

If you process with WaveHammer, of course there *will* be a change - try taking WH out of the equation completely until you track down the problem.

But I suspect what is coming out of Reason is the reason.


geoff

Subject:RE: Distortion problems abound

Reply by: JML

Date:5/25/2003 5:43:12 PM

"What do the waveforms look like fresh out of Reason ? Normalising doesn't flatten tops. "

The waveforms look fine. With Reason output set just below clipping, the waveforms come out pretty low usually. S the first thing to do is normalize them. But this brings me back to my strange problem: The waveforms look fine, but the sound is distorting. Normalization isn't clipping the waveform (as it shouldn't), but I'm still hearing distortion.

Subject:RE: Distortion problems abound

Reply by: MJhig

Date:5/25/2003 5:54:41 PM

You say you are using WaveHammer... Are you checking the next stage in WaveHammer or all of your plug-ins for over loading the next stage? In WaveHammer for example the first stage is compression (check output meters) then the next stage is Volume Maximizer (tab at bottom) where you need to check the input for clipping and so on.

Gain staging is what I'm referring to, each process starting from the start to end along the way must be checked for input signal and output signal.

MJ

Subject:RE: Distortion problems abound

Reply by: rraud

Date:5/26/2003 9:29:01 PM

In addiiton to mjhig:
Wave Hammer can be a 2 stage process. If you overload/clip the 1st stage, there is no hope for the 2cd. Make sure output level on the 1st stage is not clipping.
THIS APPLIES TO ALL PLUG-IN CHAINS.
(unless your looking for a specal effect)

Subject:RE: Distortion problems abound

Reply by: jorgensen

Date:5/29/2003 1:05:51 AM

You can overload PlugIns as much as you like without distortion - as long the following is observed.

1 - The PlugIn chainer is running in 24-bit (IEEE float) - lower right corner in the chainer window.
2 - All the PlugIns support 32-bit floating point (most does)
3 - The level is lowered at the final stage to avoid distortion when converted to fixed bit length.

You can use Wave Hammer or any other limiter to avoid the final overloading, and add some dither.

Subject:RE: Distortion problems abound

Reply by: JML

Date:5/29/2003 8:34:14 PM

This problem is also occurring when I do nothing but normalize. I had one track that I couldn't even normalize to 75% without distortion.

Subject:RE: Distortion problems abound

Reply by: JML

Date:5/29/2003 8:35:22 PM

This problem is also occurring when I do nothing but normalize. I had one track that I couldn't even normalize to 75% without distortion. I should say again, the distortion was heard, not seen.... the waveforms were just where you would expect them at 75% normalization.

Subject:RE: Distortion problems abound

Reply by: jorgensen

Date:5/30/2003 1:34:28 AM

The reply wasn't a respond to 'your' distortion problem, but the statement that overloading a plugin leads to distortion.

If you normalize to 75% with RMS you definately gets distortion.
You only allow a headroom of 2.5dB, and should AT LEAST have a 12dB headroom to avoid audible distortion.

If you do not want possible distortion/compression use Peak instead

Subject:RE: Distortion problems abound

Reply by: JML

Date:5/30/2003 8:31:36 PM

"If you do not want possible distortion/compression use Peak instead "

What do you mean? Does Peak normalize differently than SF? Is the metering different? In school I used Peak and didn't have problems with distortion in files below 0db. That's actually part of the reason I posted in the first place. When I was using Peak at school, I wasn't having any of these problems. Then, when I started using SF at home, I ran into these problems. Is SF just inferior to Peak? Not that it's an option anyway, since I'm on a PC.

Subject:RE: Distortion problems abound

Reply by: rraud

Date:5/31/2003 5:23:15 PM

Now we're confused.
Are we talking about peak and RMS db levels in the Normalize function of Sound Forge..
Or, are we talking about Bias Peak for Mac software.

Subject:RE: Distortion problems abound

Reply by: jorgensen

Date:6/1/2003 2:38:53 AM

Okay – back to the bench.

To normalize in SoundForge you can either scan the levels for the highest peak or RMS which can be considered as an average or loudness level.

In the peak mode you can only normalize to 100% and below, where the level is shifted to have the highest peak correspond to the normalize level – therefore clipping is impossible.

In the RMS mode the loudness level is set to the normalize level, and depending on the material and settings it may lead to peak clippings.
Therefore you have a lot of possibilities to change the way the scanning is done, and what to do with possible clipped peaks.
You can play with these parameters and click the Scan Levels button, below you can then see what level to normalize to without getting distortion.

Basically there is no reason to RMS normalize in SF, you are much better off using Wave Hammer or similar to optimize the loudness level.

Class dismissed.

Subject:RE: Distortion problems abound

Reply by: JML

Date:6/1/2003 6:29:18 PM

LOL. Sorry. It was a coincidence that I was using Peak at one time, so my perception of what we were talking about was, quite obviously, wrong.

OK, I have been using Peak normalization, not RMS. Even though the graphical representation of the waveforem does not indicate any clipping, I am hearing distortion. Could this be a problem outside of Soundforge, like with my soundcard?

All audio is coming from Reason, so tracking levels/distortion are not in the equation. The bounced file is within a reasonable amplitude range. I normalize using peak level, and the gain increases as you would expect. But, while the file is topping out at -2db or so, I am hearing distortion. I tried lower normalization amounts on this file, so that the peaks were even less than that, maybe -4 or -5 db, and I still hear distortion. This was most apparant in a dense string/pad track which builds a bit at the end. This was where I was hearing, clicking, pops, static, etc.

Subject:RE: Distortion problems abound

Reply by: rraud

Date:6/1/2003 7:06:44 PM

I'm not familiar with Reason.
If your just importing files, your soundcard has nothing to do with it, except what you hear on playback.
Is it staying with the same sample rate / bit leignth? A client of mine had a simular problem importing 48k files from his FAST video edit app.
Using Peak/Normalize "should" not distort them, if the the Use current scan level box is un-checked. You can however distort/clip files using RMS/Normalize.

Subject:RE: Distortion problems abound

Reply by: wedge13

Date:6/2/2003 2:55:40 AM

Not offense intended, but have you checked all of your inputs & outputs for correct orientation? I have seen instances where an output wasn't connected to the *proper input*, yet the signal managed to somehow squeak through. Of course, the signal was at that point seriously deficient in many aspects, but it wasn't always apparent until such and such happened, or the knob was turned to 11.

Cheers - Thom

Subject:RE: Distortion problems abound

Reply by: JML

Date:6/7/2003 3:52:38 PM

"Not offense intended, but have you checked all of your inputs & outputs for correct orientation?"

None taken. At this point, I'm entertaining any and all ideas. :)

Reason is a self-contained "virtual studio" application which has a few soft synths, drum machine, sample players, etc. So the audio is generated within, and stays within, the computer at all times, so incorrect signal routing is not an issue. Tracks are generated, mixed, and bounced in Reason to a stereo file, which I'm running through Soundforge for final tweaking and mastering.