CD vs MP3 Quality.

Correct me if I am wrong, but isn't MP3 a lossy compression, so regardless of the bitrate or sampling rate, you are going to lose something on the overall sound. In other words, if you have a pure analog signal coming in and you digitize it, you essentially get a digital representation which can then be encoded into a particular format (like a .WAV file for example). If the encoding is lossless, then the digital representation can be exactly reproduced (like a CD). If the encoding is lossy (like an MP3), the resulting decoded digital representation is similar, but not exactly like the original digital representation. When either of these digital signals gets converted back to analog, it is similar to (but not exactly like) the original analog signal. But arguably, the lossless representation will be closer to the original because it has more of the original information content (ie. the CD should be better than the MP3).

The point I had made in the previous thread (and I take responsibility for taking it OT, I am sorry) is that while MP3's are not necessarily as high quality as CD's (by the above argument) they are appealing because their small file size makes it possible to do more with them and the sound quality is good enough for most purposes. The ultimate point here is that quality is not necessarily the end all criterion for a lot of people.

Can you make a high quality MP3 that is as good/better than a CD? Sure. But it will be larger in size than most of your 'typical' MP3s that you see, because you can't get something from nothing.

BHoward said:

I think we should get some of the facts straight first ........Usually you will save about 1/2 the space. This is exactly the same way WinZip works...... I use 320 Variable Bit Rate MP3 encoded with the Lame encoder. I cannot tell the difference between it and the orginal on my system. .....

Click to expand...

Well, if we're gonna get facts straight, then I don't think the 2:1 compression ratio applies to what we are discussing. Try compressing a music WAV file down with WINZIP. I think it is unlikely you will see a 50% reduction. The classic 2:1 ratio that everyone uses is given that the average data entropy across all different types of files (text, database, executables, etc...). Depending on the type of data and its content, the actual compression ratio can be much different. A great example of this are computer tape drives which have a 20GB native capacity and 40GB compressed. Unless you are only backing up word documents, it is very unlikely you will get 40GB on one tape.

When talking about a digital audio file, the amount of lossless compression is based on the entropy (or repetiveness if you will) of the digital data (ie. numbers). And the data could be highly compressible, but more likely will not be very compressible at all.

With regards to your last sentence, this hits the nail on the head. Its up to the listener to decide what is best. IE. if you could encode the MP3's at 64VBR and not tell the difference, then that's what you would use, even though you know there is a loss in quality.

It should also be mentioned that the quality of D/A conversion can play a huge role in all of this. I don't know if such a thing exists, but I wonder if anyone has ever done a study comparing an analog signal, its digital representation converted to analog and then the resulting analog signal from an MP3 encoded version. It would be interesting to see what the actual error % is between the original, the digital version, and the MP3 version.