Take the 2-minute tour ×
Sound Design Stack Exchange is a question and answer site for sound engineers, producers, editors, and enthusiasts. It's 100% free, no registration required.

When a lossy audio track (of unknown quality) is imported into software and and saved as a (possibly) higher quaility after editing, what happens?

Obviously I know it can't have any more 'quality' than what it was imported with, but I was wondering if anything else happens to the track. For example does it cause unnecessary file blot? Or could it possibly make it lose quality? Or affect its quality if it is ever edited again?

Is this anything I should worry about?

share|improve this question

migrated from avp.stackexchange.com Jan 27 at 15:05

This question came from our site for engineers, producers, editors, and enthusiasts spanning the fields of video, and media creation.

2 Answers 2

up vote 6 down vote accepted

When you import a lossy audio file into editing software, that file is decoded into an audio waveform. This is a "lossless" waveform in that it's not stored in a compressed format, but it's not the same as the original waveform that your lossy file was approximating.

So there are already three concepts involved:

  1. The original waveform, which is to say, the "full-quality," unmodified waveform
  2. The lossy encoding of the original waveform (not a waveform itself, but it decodes into one that's almost the same as the original)
  3. The waveform that gets decoded from (2)

When we talk about "quality" here, we mean "how close we are to the original waveform." So waveform (1) is the highest-possible quality, and everything that approximates that involves some loss of quality.

So.

When you have your decoded waveform in your software, and you re-encode it to another lossy format, you incur a quality loss. Your new encoded file is an approximation of (3), no matter what quality you choose. But (3) was an approximation of (1). So no matter what you do, you're incurring a quality loss every time you re-encode with a lossy format, even if the amount of loss is marginally smaller (in other words, even if you choose a 'higher quality' lossy encoding than was chosen for the first encoding).

You absolutely should be worried about this if you're going to be editing and re-editing your files. If you want to preserve quality between editings you should be using a lossless format like wav, aiff, or FLAC.

You can test this by opening a lossy file and re-encoding it with something lossy, and then re-opening and re-encoding many times. It won't be long before the errors introduced are pretty obvious (though to be fair I've never run this test)

share|improve this answer
1  
A friend of mine pointed out just now that this is very similar to .JPG compression in images, and for the same reason (lossy compression). If you've ever saved a .JPG over and over, you may be familiar with the quality loss that's incurred there. –  Warrior Bob Jan 4 '12 at 20:01
2  
It's actually not quite as bad with audio codecs as it is with .jpg. There, the main problem is that part of the artifacts (the borders between the processing domains) is such a kind of signal which is not very visible to the eye but causes particularly nasty artifacts when compressed again. That's different in audio codecs: these surpress the domain borders better because they would be rather strongly audible, so these borders can't cause such a mess when you encode the file again. — Still, in principle the analogy is correct. –  leftaroundabout Jan 5 '12 at 0:30

Practically any time you convert from one lossy format to another, you will lose quality. How much quality, and what precise effect it will have depends on your encoder, and its algorithm.

It also depends on what you mean by quality. Are you referring to bitrate? Or the encoder settings used to generate the file?

It may be possible to transcode from one bitrate to another without losing any quality, but probably only if the algorithm used is specifically intended for this, and is able to notice the original compression patterns. But even then, it's probably unlikely to have no loss, and I doubt anyone has ever written an algorithm to do this--because by the time you go to that much effort, why bother transcoding at all?

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.