LibertyNT said:
I always wondered how they made MP3s smaller...
Lossy compression reduces file sizes by throwing away part of the information - that is how MP3, AAC and JPG files work. Lossless compression finds redundancy in a file and encodes that redundancy using an algorithm whose output occupies less space than the original redundancy, but it is fully reversed when the compressed file is decoded - that is how ZIP, RAR and GIF files work. Lossy compression provides greater file compression than lossless compression, but at a cost.
Try this experiment - you will need an audio workstation program of good quality such as Adobe Audition (one which can display a waveform in a mode similar to an oscilloscope trace and that can encode in mp3Pro) with a CD-ROM drive for ripping, and a high-quality listening system connected to it (I use a Tascam US-122 semi-professional USB audio interface with Sony MDR-V6 headphones that have been my reference 'phones for nearly 20 years):
Start with a music CD of known high quality, preferrably digitally recorded amd mastered with great attention to quality, and which you have frequently listened to in a high-quality listening environment. Some of you might prefer a classical music CD, but for experiments like this I prefer the title track of "Come Away With Me" by Norah Jones.
Rip the track into the audio workstation at 44.1 kHz / 16 bit, and save a copy as an uncompressed WAV file. That will be your reference. Now save it again as an MP3 file (I would suggest settings of MPEG Layer-3, 128 Kbps (MS J-Stereo) - fairly typical for audio distribution), which should yield a file about 10% of the size ot the WAV file. Minimize the audio editor, find the saved WAV file and compress it with a lossless file compressor such as WinZip or WinRAR. Note the file sizes of the WAV file, the MP3 and the ZIP or RAR file. You will see that although lossless compression may save some file size the MP3 is significantly smaller.
Now decompress the ZIP or RAR file, saving the output with a different name, and open it in the audio workstation. Invert the original WAV, sum it with the decompressed WAV, and examine the result in waveform view. Now open the MP3, sum it with the inverted original WAV and examine the result in waveform view. The waveform view in each case shows how much data was lost in the compression-decompression cycle. If your audio workstation supports one of the newer compression algorithms such as AAC give that a cycle as you did for MP3.
What you see on the workstation after summing the MP3 with the inverted WAV file is the information that MP3 throws away in order to achieve a reduction in file size. Listen to that result and you may even be able to recognize the original song or performance. Once the MP3 is created that information is forever lost. That is why MP3 may be suitable for uses where audio fidelity is not of primary concern, and is often used for commercial and some program distribution (a practice which I abhor), but is not suitable for music storage on a hard drive in a radio station where, as I have pointed out above, a $100 hard drive can hold, uncompressed, the entire library for most formats which use music-on-hard-drive.
