Almost a decade ago, I converted all my mp3 files higher than CBR 192 kbps to exactly CBR 192 kbps. This is because I can no longer here any difference between CBR 192 and CBR 320 (using and ABX test), even if I convert a particular 320 file into 192.

However a few years ago, I found that my files are just too large for me to be able to continue adding music to my playlist. So I found a new format. I am now using m4a Recently I have used the nero aac encoder with dbpoweramp. I used VBR settings set to an estimated bitrate of 160 kbps. By doing another ABX test, I found that I here no difference between my current mp3 192 and the aac vbr 160.

Now, after downloading a bunch of music, and after downloading the latest version of dpoweramp, I am shocked to find that my favorite bitrate (vbr 160) is gone. This is because dbpoweramp is now using a different encoder. FDK aac. I will have to start conducting another ABX test again to find which bitrates I will prefer for my new files.

Here is the problem, the two highest vbr quality settings of fdk aac are 224 kbps and 144 kbps. I recently listened to some songs using the 144 quality and they sound good enough. What's more, the file size got mutch mutch smaller than before.

Having these advantages, I have decided to convert all my old nero aac encoded m4a files to smaller fdk aac encoded m4a files because they have the same quality. The thing is these old files first got converted from mp3 320, then to mp3 192, then to nero aac vbr 160, and will finally converted to fdk aac vbr 144. Basically, they have been converted three times now, and will be converted again if I chose to proceed with the fdk aac conversion.

I am extremely concerned what might happen to my files while converting again and again mostly to the same formats but with different encoders because I found that I am reducing the quality by doing this. I have tested a few songs with the fdk aac encoder, they sound fine. However I have more than 2000 songs, it would be very time consuming to test them all while continuing to add new files.

Is this a good thing to do? Should I proceed to converting my nero aac encoded files to fdk aac because the difference aren't noticeable? Or should I download the original sources for all my 2000 files and reencode from there?

  • 2
    Not really an answer to your direct question but: Have you considered just getting more storage? 4TB drives are pretty cheap. Even at 50MB for each file, that's 75,000+ files. – Ouroborus Nov 10 '16 at 5:12

I agree with Seth:

you'll have to decide whenever the degradation in quality is acceptable for you.

However there's one big flaw in your method. Every time you made a test ("I hear no difference") you compared files that are quite close in the compression/degradation chain. Therefore you are prone to sorites paradox.

Or maybe you are a boiling frog? :D

If you had many cups of coffee and the next one had one grain of sugar more in it than the previous one, then you couldn't tell the difference in taste between two consecutive cups. Still there would be a huge difference between the initial sugarless coffee and e.g. 2345th one. This example is exaggerated but you get the idea.

My practical hint is: make tests, starting from a high quality file, recreating your chain of compression. Compare to the original file and then decide how much degradation in quality is acceptable.

Let's say you will have:

  • original file A, a wav ripped from CD audio;
  • file B, mp3, CBR 320, made by compressing A;
  • file C, mp3, CBR 192, made by recompressing B;
  • file D, aac, VBR 160, made by recompressing C;
  • file E, aac, VBR 144, made by recompressing D.

Compare C, D, E to A (or at least to B). In my opinion this is The Right Way to test these things. But beware, it may be the red pill for you. I won't be surprised if you suddenly find your current level (D) unacceptable.

  • that sounds depressing. Regarding tests, I usually picked one death metal song (because it's musically complex and I believe a better way to test music) and judge the quality of the bitrates soli based on this particular track. If I were to find "e" is acceptable, is it possible that the quality is acceptable only on some of the songs and probably not on all songs? – morbidCode Nov 11 '16 at 15:00
  • @morbidCode I guess it is possible. In particular you may find e.g. Justin Bieber's songs quality less unacceptable when the bitrate approaches zero. ;) Jokes aside: I think it's wise to test with a complex song because of its entropy (there are more "details" to be lost). I'm no expert in audio processing nor a musician. I just know the sorites paradox may appear in many fields, also the information theory applications are wide enough to make some general qualitative estimations here. Your impressions will depend on your hearing, age, hardware, musical taste. In the end you will be the judge. – Kamil Maciorowski Nov 11 '16 at 18:32
  • as you predicted, the results are quite depressing. I tested a flac file, then converted into 320 mp3, 192 mp3, 160 vbr aac and finally 144 aac. From 192 and above the difference is really really small, but I can still recognize them. Nothing is completely as good as the flac file. I can even hear the little difference between mp3 320 and the original flac file. I've never noticed this before. I am beginning to hate lossless formats now. – morbidCode Nov 13 '16 at 15:30

Yes each lossy compression will lower the quality. Converting a file multiple times using the same lossy compression will usually lead to worse output each time. You already noticed it yourself from your own question:

I found that I am reducing the quality by doing this.

If you still have access to the original source of those 2000 files why not just use that source in order to listen to them? Popular streaming services could also help you out in order to be able to listen to those songs without actually downloading them.

What Ouroborus suggested, getting more storage, is probably the better solution. That way you'd only "need" to convert them once or not at all, as current storage and file sizes are pretty big/small. In addition that way, assuming it's a good source, those files only got compressed once which would

In the ended you'll have to decide whenever the degradation in quality is acceptable for you. Depending on it you could go either way.

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.