0

From time to time I have some files which I need to transcode for my personal mp3 player to play it.

They may be recordings in other formats or even videos.

I like to use VBR, as it gets the best quality/size ratio, but much of the times I really don't really know the input audio quality, so I don't know what quality to use - I don't want to use VBR0 for a poor quality input (and waste space) nor do I want to use a low output quality for good recordings.

So, the question is: how can I know what VBR quality would best fit the source file's quality?

Is there a way to know what is the highest MP3 bitrate that a file would need to [almost] not loose quality?

I've heard about things like frequencies analysing, but really don't know what they are nor how to do them, so if that's the way, I appreciate any further info.

  • This appears to be off topic as it is about personal music compression rather than sound design. Further there really is no "best" option, there is simply a scale of trade off between size and quality. – AJ Henderson Apr 4 '14 at 13:57
  • @AJHenderson: sorry about that. Do you know any SE site that this question suits better? – mgarciaisaia Apr 4 '14 at 17:06
  • @mgarciasaia - maybe superuser, though really, it is probably either too broad or too opinion oriented no matter where you put it. "Best" is not an objective measure and for some people the smallest files size will be best, for others, nothing short of full quality 24 bit FLAC is best. – AJ Henderson Apr 4 '14 at 17:07
  • @AJHenderson: sorry, hit Enter by mistake. About the "best" thing, I supposed there's a top quality above which the file size grows but the quality remains the same (as the output quality supersedes the source quality. By the comments, it seems that's not that true, but the "best" criteria I was targeting was an objective one. Thanks for your critics. – mgarciaisaia Apr 4 '14 at 17:09
  • @mgarciasisaia - ah, gotcha, the term for that is "lossless" Any recompression that isn't to a lossless format will have further loss with very rare exception (there are some lossy formats that will allow a file that isn't modified to be re-encoded without further loss if it is the same file type, but I'm not even sure there are any of those for audio.) – AJ Henderson Apr 4 '14 at 17:31
2

A transcode will always lose quality. If your source files are already compressed with a lossy codec, then that process will have added a load of noise to the audio - the genius in good codecs is that they can make the noise un-noticeable by carefully selecting where (in the spectrum) and when to insert the noise.

Re-transcoding this compressed file, the new codec now has to try and encode all the additional noise - it doesn't know which bits of the signal are the wanted sound, and which bits are compression artefacts. Noise is much harder to compress than correlated sound, so most of the output bitrate will be trying to accurately reproduce the noise, and only a small proportion will be used for the wanted signal.

So, regardless of your input bitrate: the higher the output bitrate, the better the output quality will be.

If you want to use the input bitrate as a guide for your output bitrate (which would probably be appropriate - you don't want to waste disk space on files which are already very low quality), you could use a tool like MediaInfo to check the source bitrate (could be done programmatically if required), and then use an output bitrate proportional to that.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.