MP3 does result in lossy compression–but is that a bad thing? Feedback hed: Don’t believe everything you hear dek: MP3 does result in lossy compression–but is that a bad thing?
Reader Gordon Kion III wrote claiming that MP3s had CD-quality sound, and tried to back it up with an experiment that you could do yourself. While the experiment will produce the file sizes that Kion suggests, it does not prove that MP3s are CD quality.
MP3 compression is not like ZIP compression. MP3 uses lossy compression, while ZIP is lossless. In a nutshell, the idea behind MP3 is to remove parts of the audio signal that the majority of people would not notice, based on a psycho-acoustic algorithm. This removes quite a lot of data. The amount of reduction is based on the bit rate at which you encode a file. The lower the bit rate, the lower the audio quality. The idea is similar to JPEG compression: The smaller the file size, the worse it looks.
The reason that Kion’s experiment will produce an uncompressed .wav file of large size is due to the nature of uncompressed Pulse Code Modulation files (raw digital audio). .Wav files store as much sound data as their sample rate will allow. So when you convert an MP3 to a .wav file, that .wav file will write zeroes in the places that the MP3 format had skipped. A .wav file sounds like its source; MP3s do not. Take a low-quality MP3 file and convert it to .wav; it’ll still sound poor. -Eli Kae Moore, e-mail withheld upon request
Gordon Kion’s letter and Dan Heilman’s response miss the point about MP3. While Kion’s anecdotal arguments hit the mark, his technical arguments are way off base. The fact is, MP3 always was a lossy compression technique. It was designed that way. It uses perceptual encoding techniques to throw away information not considered to be audible under casual listening conditions. It certainly does a good job, particularly at the higher bit rates, since most people do not notice a significant difference between a CD and its MP3 counterpart.
But the music content of an MP3 encoded file is not the same as its original source. Kion’s argument about the size of a .wav file doesn’t work, because the size of a .wav file depends on time, not content. A 60-second recording of silence and a 60-second recording of any music would result in a .wav file of the same size.
I would suggest that Kion perform the following experiment: Obtain a clean .wav file from an original, uncompressed, source. Compress the file using MP3. Next, uncompress the MP3 file into a new .wav file. Finally, compare the original .wav file with the new .wav file, byte by byte. You will find that the two files will not be the same. (I would also surmise that if you continued to encode and decode in and out of MP3, the sound quality would begin to degrade considerably.) The difference is due to the losses in MP3 compression. However, while the files are different, you may not hear a difference. So while MP3 files are not CD quality, they are usually good enough.
The fact that Heilman can hear a difference is not an indictment of MP3. If I sit down and really concentrate, I notice differences as well, but so what? I am amazed at how much compression they are able to accomplish and still produce a listenable product. The same goes for DVDs, which use a lossy video compression technique. It’s not the same as watching a pristine 70mm print in a state-of-the-art theatre, but pretty darn good for the home. -Nicholas Galemmo, e-mail withheld by request
To start a discussion or ask a question, e-mail firstname.lastname@example.org. Letters may be edited for style, length, or content. No anonymous letters will be published.