Lossy to lossy conversion via wav

[skypilotpete]

If I am converting from one lossy format to another (say, .aac to .mp3) I have been in the habit of first converting to .wav, and then converting the .wav files to .mp3. My thinking has been that converting to .wav is simply a process of decompression, and should not involve any loss of quality. Then (so my uneducated thinking goes) converting the uncompressed .wav file to .mp3 should be a simpler and more straight forward process for dBpoweramp than a direct conversion from one lossy format to another, and MAY give better results.

Is there any validity at all in this thinking, or am I deluding myself? Is it just as good, or better, to convert directly from one lossy format to another (always assuming that the conversion HAS to be done!)

Also, I have recently read some comments on another forum which claim that converting from lossy format to wav is NOT a lossless process and may introduce some artefacts. Does anyone here know about this?

[mville]

Re: Lossy to lossy conversion via wav

Ideally, you would only convert from a lossless format to a lossy format.

If you have to convert from lossy to lossy, you will lose quality.

Do not convert lossy to wav to lossy as you gain nothing and more likely get less quality.

[Porcus]

Re: Lossy to lossy conversion via wav

mville is right. Avoid transcoding - even though you will often not hear the difference, it is unnecessary except for e.g. compatibility.

You can re-encapsulate an audio stream in a different container without transcoding, provided that this container supports the stream in question (for example, you can "convert" a raw .aac file to an mp4 file with precisely the same audio, using software that just puts an mp4 container around the bitstream without actually decoding it). But aac to mp3 is a no-go. Don't do it. And if you do out of, say, compatibility reasons, then keep the original.

Now what you have heard that "converting from lossy to wav is NOT a lossless process", that is for all practical purposes wrong; e.g. mp3 is specified in terms of decoding, so an mp3 file contains a unique decoded signal (up to roundoff errors). But PCM in WAV is not a floating-point format, so technically speaking, you will get the signal truncated somewhere just like you will when you playback it through a DAC.