Does lossy to lossless make sense in this case?


So, from my understanding audio lossy to lossless conversion is a no no. However, what if I want to edit a lossy file (e.g. change tempo, make cuts etc) and I don't want to lose quality by converting it from lossy to lossy, would converting it to lossless, editing it and saving it as lossless be an acceptable solution?


Posted 2017-03-10T16:09:43.613

Reputation: 23

3Yes. Not only does it make sense, but it's what I'd recommend. The issue isn't so much lossy to lossless (you've already lost what you've lost, now let's not lose any more), but rather people who think that they can recover lost information by converting lossy to lossless. – Linuxios – 2017-03-10T17:13:45.827

1@Linuxios - flesh that out a bit & it's the answer. – Tetsujin – 2017-03-10T17:38:09.240

@Linuxios Awesome, that's what I thought. Thanks! – Wyse – 2017-03-10T18:38:26.893



It's sort of the best you can do. "Make cuts" could likely even recompress reasonably without further quality loss if you manage to make the cut exactly in units of the compressor's frame size. Change tempo is not likely to recompress without further loss.

For JPEG compressed images there are programs which can do some basic operations like rotating, mirroring, and some cutting without inducing further loss: they work on the compression coefficient matrices directly.

I am not aware of anything similar available for audio, and the kind of operations you want to do here also do not seem to map well to that model apart from cutting.

So practically speaking, your path is the best for avoiding further loss but it gives you lossy quality material at nonlossy quality file sizes.

Also if you do editing to clean matters up (like equalizing single instruments), the assumptions of the compression algorithm about what compression artifacts will be inaudible may become wrong. Which is the reason that lossy compression should always be only the very last step before a production is given to the consumer. That gives the best compromise between quality impact and file size reduction.

From the examples you cite "tempo change" is such a step: compression algorithms can assume that certain artifacts are hardly detectable if they are transitory and over fast. Extreme tempo changes could violate that assumption.

But at any rate: you cannot do better than you were planning to do.


Posted 2017-03-10T16:09:43.613



When processing audio data it is the process of "encoding" that causes loss of audio data. So when encoding with a "lossy" codec, it's the process of "encoding" that causes the loss - not "decoding". Conversion from "lossy" to "lossless" therefore will not cause any additional loss of data as the "loss" has already occurred.


Posted 2017-03-10T16:09:43.613

Reputation: 7 535