So I was reading this article (actually, someone's comment) on what converting from a lossless codec to mp3 does to your sound. I wanted to test the hypothesis: What does mp3 add (or remove) from the fundamental lossless recording?

I took my own song, "Rain", and put both the WAV and MP3 in Audacity. Here's how it looked:

    nothing in the top one, but some strange compression in the bottom one...

From there, I uploaded it to Sound Cloud. Here's what I got:
https://soundcloud.com/davdud101/rain-mp3-analysis
    notice how all you hear is strange noise from the drumkit and vox

So what say you? Is this a major degradation of the sound quality?

MP3s are meant to be compact so it has to do something to the original waves to achieve the smaller size. That's why it's important to have a good program to do the conversion. If it compresses it in a crummy way you might hear it more. Just like there are different algorithms for dumping 24/32 bit files to 16 bit. The good news is MP3s have been around for a long time so there are many programs that make the final product almost imperceptible from the original to the human ear. If you look at it though as you have there you will see a difference. If you are unhappy with the quality try another program to see if that helps. The important thing is what is perceived by the listener. If things are mixed, mastered and converted properly the average music lover or even a seasoned engineer won't be able to notice much difference.

It depends on a few factors too such as the bit rate in mp3 encoding. Taking the bit rate up to 256 or 320 KBits/sec will give you a better sound too. You don't want to mp3 encode a very heavily mastered wave file either that is slammin right up to -0.1 dB FS. You almost need to master slightly differently or at least limit to say -1 dB FS instead. There are a few techniques to getting a good mp3 version of your mix. Not all mp3 encoders are created equal either. 

But get into AAC encoding instead here:

http://en.wikipedia.org/wiki/Advanced_Audio_Coding

The successor to mp3. It is already way ahead and the results are quite spectacular. A much more interesting format.

Unfortunately these compressed formats are becoming all too common as the main distribution product and listening media. It is still important to work in higher resolution formats such as we do in order to at least have a great mix to convert to a lower res if need be.

I find compressed formats great for sending mixes and things to clients so they can preview things. In that mode they are amazing and very helpful. Well mixed and mastered wave files still sound better to me.

Here's how you can get an idea of what the MP3 encoding is removing. 

Put a single hi-hat hit at the front of your song in the count-in area. This will be used later for alignment. Export the project as a normal wave, then create an MP3 from that wave. Now open a new project and import both the wave file and the MP3. Using the hi-hat hit, visually line the two files up. Match their levels to within a fraction of a db (this will take some trial and error). Now invert the polarity of either of them and bounce the two to a third track. If you've matched levels well, this third track will contain pretty much what the MP3 encoder has removed. 
 
I should point out, though, that this is purely an intellectual exercise and has no practical value. The whole concept behind perceptual encoding is that what's being removed is stuff you can't hear anyway. So just because you can hear it in isolation doesn't mean it was audible in the original mix.

MP3's first remove a great deal of dynamic contrast.  Sometimes it is not noticeable with loud heavily compressed music, but any kind of music that relies on sharp transient responses among the entire frequency range will suffer.  I have recorded plenty of chamber music that becomes lifeless once converted to MP3.  Of course on most consumer systems and headphones, the differences are less pronounced. 

well..... this is some interesting stuff....

I don't pretend to know the intimate details of what happens behind the curtain .... but I do know that I always try to use the highest bit rate on the MP3 that will fit into the particular mp3 website's player given their limitations on file size. I shoot for 320 but will back off as needed. 

320 on my portable players always. 

I also know that something is missing because a wave of 40MB gets slammed back to 4 to 5 MB in the mp3 format..... so it had to drop something..... mathematically, it appears to be a ..... keep one bit, drop 8 or 9.... keep one,,,, drop another 8 or 9.... but I know it's not that simple.

I know the basic rules and pretty much follow them..... leave head room, use the highest bit rate possible, etc....

Have you guys tried this MP3 bit-rate blind ABX? It's an eye-opener.

I was feeling pretty cocky after getting 5 out of 5 correct, even if that wasn't statistically significant. By the time I'd done 10 tests, my percentage was down to 83%, and after 15 tests it was 71%. It depended heavily on the style of music being A/B'd. I could nail the Jazz samples every time, but with distorted guitars and wailing vocals it was a total guess.

davdud101


So I was reading this article (actually, someone's comment) on what converting from a lossless codec to mp3 does to your sound. I wanted to test the hypothesis: What does mp3 add (or remove) from the fundamental lossless recording?

In theory lossy compression essentially tries to replace the parts of your audio that you are unable to perceive with noise.

At higher bit rates it works very effectively. At lower bit rates the encoders start to have to remove/replace some stuff that is not perceptually masked and artifacts can become apparent. AAC is better at lower bit rates than MP3.

Note that "what happens" depends on the signal present at any given time, the lossy encoder, the bit rate, etc. - so it's hard to generalize too much. 

It's also very hard to gain any useful knowledge from looking at waveforms or listening to difference signals since it completely ignores the perceptual masking that is at the very core of lossy compression. Because if you can't perceive the stuff that's getting thrown away, it doesn't matter what or how much of it there is.

 To say it "adds" something is totally opposite. As others have already mentioned Mp3 removes large portions of an audio file to make it smaller. The whole idea behind all of these formats isn't improved sound quality. The goal is to remove as much of the file as possible without any noticed difference. With a goal like that it isn't any wonder than anyone who wants to make things sound better might have a gripe with the format.
  Even if we can't always hear a difference, just knowing that the file has been chopped is enough to make any audiophile worth his weight want to run the other way. FWIW you can upload to Soundcloud in high bitrate wave files. This will take more of your storage space but the files will sound better. On my end it sounds like a rhythm track but one missing a lot of umphh.

 At best Mp3 and any other lossy format are a compromise to the sound in order to make it more portable.


 Scientists have figured out the ranges where we hear the best and removed the rest...but in some material you can tell a difference.

 What was that file you uploaded. Is it a differential between tracks like Bit mentioned. Did it sound like that before you uploaded it?

It's actually the exact process Bit said- took an mp3 and a WAV an inverted one. It's definitively an interesting topic to study, but with what you guys say, all there's is to know is pretty simple; removing as much data without much noticeable change in sound quality.
In any case, it's still really cool to try.