Question about Digital Peaks...

  • Thread starter Thread starter Uladine
  • Start date Start date
Uladine

Uladine

New member
Okay, I know digital peaks should be avoided at all costs, they sound awful, and they are seen as errors at a CD duplication plant. But what I was wondering is if the peak from an initial guitar track or something else that is inaudible will carry through to the final mix on a master CD.

For example, say you recorded a guitar track and you failed to catch a minor peak that is inaudible and dosen't alter the sound of the track, then you digitally mix down all your tracks to a stereo pair that has no peaks in it, then burn that stereo pair to a CD. Is there still an error there? I've been very curious about this for the longest time.
 
I'm not an expert here, but...

First of all, I think you mean clips, not peaks. All waves have peaks, but they shouldn't clip.

Secondly, I would suggest that if you can't hear it, it probably ain't clipping.

Lastly, I think clipping is defined as 3 consecutive samples that exceed 0 db. So if you have a single sample peak that is exceeding 0db, it wouldn't be clipping.

I forget where I picked up this last piece of infomation. If it's wrong, I'm sure someone will come along and let me know. :)
 
Just to clarify that last point, there is no such thing as a sample over 0 dB. If the convereted signal goes over zero dB, all the samples will have the maximum possible value -- a flat top to the waveform. So the detail of the actual sound is completely lost and you get an unnatural waveform and it sounds awful. I don't know what the three samples thing is... unless that's the threshold when it becomes discernable (but even three max-value samples in a row would only be a tiny instant -- at 44.1 kHz, that would be 3/44,100ths of a second, pretty damn short). But I'm willing to believe that a three-sample plateau might create an audible artifact... our ears are much sharper than our eyes in this regard (30 sequential still pictures per second is enough to fool our eyes into seeing smooth motion).
 
And just to clarify a bit furthur, CD plants don't see clipped off peaks as errors. They can't distinguish between an intentional 0db and a 0db that was created due to clipping.

Slackmaster 2000
 
Just to clarify even more.

It is actually 8 samples, and sometimes 12 samples if the frequency is high enough before the ear actually hears the digital distortion. This info came via something I read by Bob Katz somewhere, so I would tend to favor it as the truth. I have many recordings where I know damn well I had "overs" and didn't hear any distortion, nor any bad artifacts in the sound. I suppose the quality of the converter, and the quality of the dithering at the A/D conversion would have something to do with how well overs are handled.

And true, dupe plants have no way of knowing what was intentional and what was not! The errors they look for are called Block Rate Errors, and those are a result of the actual burning process, and have NOTHING to do with the .wav/.aif file that was used when burning. To further this, I have analyzed several modern CD's where 0dB was achieved quite often where there is NO audible distortion (and even a few where there IS!!! but I can only hear those distortions with my 24 bit soundcard. I NEVER hear them on a CD player.....)

I once recorded a guitar solo on Type I ADAT's where I pegged the crap out of the A/D. Digital distortion was totally evident, BUT, it actually in this case sounded rather good! :) It was an intentional thing and the effect was exactly what I wanted for the sound. So, digital distortion CAN be used creatively too, although I wouldn't recommend it as a common production technique...;)

Yes, you should try to avoid "overs" during the A/D process, but, if you have one, take a good close listen to the track. If you have a decent monitoring system and good D/A converters, and you don't hear distortion, don't worry about it! You either didn't have an "over" or the "over" was short enough to where your ear couldn't detect digital distortion.

Good luck!

Ed
 
So basically if you cant hear it dont sweat it? I can't remember where I heard the thing about the clip producing errors at the cd plant.

Sorry I meant clip, not peak, in the original post. Thanks for correcting me.
 
You must log in or register to reply here.
Back
Top