Skip to main content
Topic: Compressing an already lossy compressed audio (Read 1092 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Compressing an already lossy compressed audio

Hello, sorry for noob question.
What i would like to know is if converting flac>mp3 320>mp3 256 would be worse than just converting flac>mp3 256?
I'm asking this because i have heard that, for example, converting mp3 320 to opus 128 is a bad idea, since that way you are putting sound through two different psyhoacoustic models and also compressing an already lossy compressed audio. But i wonder if doing an mp3 320>mp3 256 is similarly bad or not so much, since in this case you are dealing with the same psyhoacoustic model.

Re: Compressing an already lossy compressed audio

Reply #1
If you have the original FLAC file you go FLAC > MP3 256 kbps but if all you have is MP3 320 kbps than you're stuck with two possibilities such as keep the original file as is or convert MP3 320 kbps > MP3 256 kbps.  OPUS is a different format but depending on what you choice in the encoder it'll either sound the same as the original MP3 (best case) or worse than the original MP3 (worst case).  There's no way of knowing what the best settings are from converting MP3 to OPUS for a particular audio recording and application is but you can listen to the results and find one that sounds fine to you.  The same thing could be said for MP3 to MP3 conversion, it either sounds the same as the original or worse than the original.

If you truly have the original recording in FLAC than I use that instead to convert it to MP3 256 kbps or OPUS 128 kbps.  It's generally considered a bad idea to do lossy to lossy conversion because it's pretty much a roll of the dice, you could end up with something that's worse audibly than if you simply used a lossless source.  That said some lossy formats do handle transcoding better than others.


Re: Compressing an already lossy compressed audio

Reply #3
Check my very old reencoding listening test:
https://hydrogenaud.io/index.php/topic,32440.0.html

At this time, LAME to LAME performance wasn't very good. Encoding from a lossy format to another sounded better (and probably very close to proper encoding from original lossless source).

Re: Compressing an already lossy compressed audio

Reply #4
If all you want is to save a little space from your CBR320 mp3's then you should look at mp3packer. That can losslessly convert the file to a somewhat smaller VBR file.

Re: Compressing an already lossy compressed audio

Reply #5
mp3packer is good for 2-10% savings (typically; it depends on the file), but the OP may be looking for bigger savings.

Of course, you want to compress from an original source, not just because of different formats but because of echoing, artifacting and other issues with lossy compression in general. That aside, the best formats to transcode are AAC and Opus, according to that 100-pass recompression test mentioned by lvqcl. (The actual reason why Opus didn't turn out well in that test was the 48KHz sample rate conversion not the transcoding. YouTube has plenty of songs originally uploaded in MP3 format that if the bit rate was high enough it gets played back in Opus instead.)

Still, transcoding will lose quality no matter the bit rate.

Re: Compressing an already lossy compressed audio

Reply #6
Where can I find mp3packer? All links that direct to mp3packer are not working (at least the ones I found on the internet)

Re: Compressing an already lossy compressed audio

Reply #7
Hi is a link to the gui which includes mp3packer.exe

Re: Compressing an already lossy compressed audio

Reply #8
The HA Wiki download link for MP3Packer still works. (7Zip archive.)

 

Re: Compressing an already lossy compressed audio

Reply #9
Thanks a lot. ;)

Re: Compressing an already lossy compressed audio

Reply #10
Thank you, guys, for all the answers and links.
(The actual reason why Opus didn't turn out well in that test was the 48KHz sample rate conversion not the transcoding. YouTube has plenty of songs originally uploaded in MP3 format that if the bit rate was high enough it gets played back in Opus instead.)
Can you elaborate on that please? As far as i know, mp3 can be 48kHz too, so what was the issue in that particular case you provided? Do you mean the original track(that was uploaded to youtube) was 44.1kHz for sure?

Re: Compressing an already lossy compressed audio

Reply #11
I would imagine the MP3 and AAC files of commercial songs uploaded to YouTube would typically be 44.1KHz, given the common sources—downloads and CD rips. MP3 supports 48K, but AFAIK it's pretty uncommon. MP3 by its nature is inefficient at storing bands 16KHz and above, and good encoders tend to cut off at 20KHz by default to prioritize preserving the more audible content.

It probably doesn't matter much what the sample rate was since a single resampling and transcode to Opus isn't going to affect the integrity of the sound much. The codec is very efficient, and YouTube Opus streams tend to average above 128 kbps.

Re: Compressing an already lossy compressed audio

Reply #12
I would imagine the MP3 and AAC files of commercial songs uploaded to YouTube would typically be 44.1KHz, given the common sources—downloads and CD rips. MP3 supports 48K, but AFAIK it's pretty uncommon. MP3 by its nature is inefficient at storing bands 16KHz and above, and good encoders tend to cut off at 20KHz by default to prioritize preserving the more audible content.

It probably doesn't matter much what the sample rate was since a single resampling and transcode to Opus isn't going to affect the integrity of the sound much. The codec is very efficient, and YouTube Opus streams tend to average above 128 kbps.

Good MP3 encoders want nothing to do with sfb21 to begin with. I think even the -Y switch for LAME is too much.

AAC can handle high frequencies efficiently, but FDK-AAC doesn't actually use it. It saves the bits to be used on more important stuff.

Apple AAC is so incredibly wasteful, at least the iTunes Plus files are, that it's actually the hardest to distinguish from FLAC. And that's not a compliment. They're just wasting space.

Re: Compressing an already lossy compressed audio

Reply #13
If you have the original FLAC file you go FLAC > MP3 256 kbps but if all you have is MP3 320 kbps than you're stuck with two possibilities such as keep the original file as is or convert MP3 320 kbps > MP3 256 kbps.  OPUS is a different format but depending on what you choice in the encoder it'll either sound the same as the original MP3 (best case) or worse than the original MP3 (worst case).  There's no way of knowing what the best settings are from converting MP3 to OPUS for a particular audio recording and application is but you can listen to the results and find one that sounds fine to you.  The same thing could be said for MP3 to MP3 conversion, it either sounds the same as the original or worse than the original.

If you truly have the original recording in FLAC than I use that instead to convert it to MP3 256 kbps or OPUS 128 kbps.  It's generally considered a bad idea to do lossy to lossy conversion because it's pretty much a roll of the dice, you could end up with something that's worse audibly than if you simply used a lossless source.  That said some lossy formats do handle transcoding better than others.

If all you have is 320 CBR, you can get closer to 256k by running MP3Packer on them without any further loss in quality as long as the thing you're playing them on in the end is okay with VBR files.

 
SimplePortal 1.0.0 RC1 © 2008-2019