S3/Diamond Rio 600 Review

Everyone knows how large audio files can get; increasing the quality and audio accuracy of a recording increases file sizes, too. This is exactly the reason devices such as the Rio make use of compressed formats. These compression technologies use advanced techniques to reduce the size of an audio file considerably--but how?

MP3 is the famous and widely used audio compression standard. MPEG itself stands for ‘Moving Picture Experts Group,’ the official title of the committee formed under ISO (International Standards Organization). Another point to note is that what we call MP3 is really the audio compression algorithms of the MPEG1 standard, the 3 in MP3 refers to the layer 3 of MPEG1. Each of these MPEG1 audio compression layers is more complex than the last, while also giving better quality audio. Layer 1 produces a typical 1:4 compression ratio, with layer 2 providing an improved 1:6 to 1:8 ratio and finally the layer 3 giving 1:10 to 1:12 compression rates.

Without compression audio files sizes are enormous: 44.1 kHZ 16-bit Stereo audio produces 176,400 bytes of data a second. This works out at a giant 10,584,000 bytes every minute, which is roughly 10 MB. This is why a 650 MB CD-ROM audio disc holds about 74 minutes of audio, because it is stored in this uncompressed format.

How can these file sizes be reduced? The first approach is to sample less information, resulting in less data being stored and smaller file sizes. The sampling frequency of a recording determines the accuracy of the signal recording. When the sampling frequency is reduced, so is the precision of the recorded audio. This loss of accuracy is clearly evident to anybody listening to samples taken at different sampling frequencies.

There is a relationship between sampling frequency and frequency response that is explained by the Shannon-Nyquist Theorem. Basically this theory states that for an audio signal to be reproduced using PCM, at least two samples of each cycle of its waveform must be taken. This results in the highest frequency that can correctly be recorded is half the sampling rate used. This somewhat strange idea is known at the Nyquist Limit. Upon realization this means that a CD audio system may use a sampling frequency of 44.1 kHz, but this results in frequencies being played at up to 22.05 kHz. At first this may sound strange, but when you remember that humans are incapable of hearing anything much past that range, it becomes clear that such a frequency is acceptable for human listening.

The way MP3 tackles the problem of file size is not by storing less data, but by encoding. JPEG stores images using a lossy compression method, meaning data is lost in the storage process. When a file is saved as JPEG, the first pieces of information to go are parts of the image that can’t be seen or noticed by the human eye. MP3 compression follows the same approach: if a human can’t tell the difference between a piece of data being there and being removed, then it can be removed.