MP3 cutoff frequencies

The quality of an MP3 depends largely on the encoder. The specifications, in fact, tell you what to do, but not how to do it (unlike decoding, which is a purely mechanical process). Precisely for this reason, a low-quality encoder is recognizable even when listening to a 320 kbps track. It follows that it makes no sense to talk about the listening quality of a 128 kbps or 192 kbps track without referring to the codec used. A good 128 kbps MP3 encoded by a good encoder produces better sound than a 192 kbps MP3 encoded with a poor encoder.

The tests, like the one in the previous post, are performed with L.A.M.E. (Lame Ain't MP3 Encoder), which is recognized as one of the best (perhaps the best for compression from 128 bits and up).

LAME itself provides the cutoff frequencies for various compression levels. During the encoding process, a series of filters are activated to divide the signal into bands, stopping at a different level for each bitrate. The portion of the signal exceeding the last band is eliminated with a low-pass filter that begins attenuating at a certain frequency (start in the table) and cuts completely beyond a certain level (end in the table).

It's also possible to disable this filter (there's an option in LAME), but doing so risks producing artifacts that are usually identifiable as a certain type of effect that resembles a flanger (often heard in the audio of ripped and overly compressed films).

Here's the table:

kbps	*cutting area: start, end*
128	16538 Hz – 17071 Hz
160	17249 Hz – 17782 Hz
192	18671 Hz – 19205 Hz
224	19383 Hz – 19916 Hz
256	19383 Hz – 19916 Hz
320	20094 Hz – 20627 Hz

The table reads like this: for example, at 128 kbps, frequency attenuation begins at 16,538 Hz and increases to 17,071 Hz, after which everything is eliminated. Therefore, theoretically, even a 320 kHz MP3 is significantly inferior to CD quality. In reality, frequencies above 20 kHz are inaudible, but appreciating a recording is often a matter of listening education:

A test given to new students by Stanford University Music Professor Jonathan Berger showed that student preference for MP3 quality music has risen each year. Berger said the students seem to prefer the 'sizzle' sounds that MP3s bring to music.[27]
Others have reached the same conclusion, and some record producers have begun to mix music specifically to be heard on iPods and mobile phones.[28]
However, the study was criticized for being a short-term A/B test, which does not reflect the listeners preferences when they listen to music for prolonged periods.[29]
[wikipedia]

[27][28][29] are bibliographic references cited in Wikipedia. Click on the numbers to go to the articles. The first is Berger's article, the others are comments. I think we should begin a serious reflection on the changes in the ways we listen to both music and natural sounds.

Back