Does "converting" mp3 file two times, makes a difference?

Question

I am interested in: I have Digital Voice Recorder appliance, which saves audio in 192kbps MP3 file. I put that file in AudaCity, then I export the final (optimized) mp3 file (192kbs), which i put in Adobe-Premiere and then re-exporting that video (which again encodes audio file, with same 192 kbs bitrate).

I am interested, will be there any difference between the exported sound (from audacity) and in the sound of video file (because that sound was again encoded)?

will there be any significant difference?

Answer 1

Yes, there will be a difference. Mp3 is a lossy codec that deteriorates with every generation. It's bad practice to use mp3 (or lossy compression in general) at any stage where the audio may yet be edited or reencoded. Some programs are smart enough to simply copy the original audio stream without reencoding if possible, but I can't say whether this applies to your workflow.

As for whether or not the difference is significant, that of course depends on your expectations of audio quality and the contents of the file. You can determine this yourself by comparing the 'original' mp3 to one that has been reencoded twice. Results will vary based on the source material, but for a quick impression, you might find this experiment interesting.

Answer 2

It very much depends on the encoder/decoder. Intuitively, the uncompressed data matches the assumptions of the encoder perfectly and there is no reason to throw away more information since the fixed bitrate can be met.

However, I seem to remember that ATRAC (the minidisc encoding) was explicitly designed to recompress badly in order to degrade any kind of copying (analog or digital via S/PDIF) in addition to the digital copy protection schemes.

Also, there are obviously some things you do in Audacity ("final"/"optimized") that are not pass through, and also in Adobe-Premiere. Even if you only do video cutting, the audio framing will most likely be disturbed (I doubt that independent audio frames are a clean fraction of video frames), causing an entirely new compression ballgame.

If you have any chance to do that, I'd not let Adobe-Premiere reencode but rather repackage the existing MP3. Alternatively, record in WAV. Or at the very least, do all your processing in WAV and only recompress for the final exported video.

Answer 3

Your original workflow compresses 3 generations. At the recorder, in Audacity, and in Premiere.

The best way would be to record in WAV, edit in Audacity, export to WAV, import that file in Premiere. You will then, in effect, only encode to MP3 once.

That said, unless you have a recording studio or controlled environment with very low natural echo/reflection, low noise, a quality microphone, pop filter, wind screen, etc, then none of this is of great consequence. I've produced MP3s recorded with a $20 microphone that sound better than a $250 microphone because I was able to control the recording environment. In summary, I don't think 2 or even 3 passes will do too much damage if dealing in uncompressed WAV is too much of a hassle.