Non-uniform quantization reduces audible noise by spacing levels by probability.

Evidence

One solution to this is to use non-uniform quantization, sometimes called non-linear quantization, in which the levels are not equally spaced. By having the less probable higher levels further apart, the increased quantization error that results is more likely to be masked by the signal, and, because it’s less probable, the average signal to noise level does not increase. In fact, for a given number of levels the average signal to noise reduces because the more likely levels are closer together and thus generate less quantization noise. — David Howard, Jamie Angus, coustics and Psychoacoustics, 2662

By having the less probable higher levels further apart, the increased quantization error that results is more likely to be masked by the signal, and, because it’s less probable, the average signal to noise level does not increase. In fact, for a given number of levels the average signal to noise reduces because the more likely levels are closer together and thus generate less quantization noise. — David Howard, Jamie Angus, coustics and Psychoacoustics, 2662

One solution to this is to use nonuniform quantization, sometimes called nonlinear quantization, in which the levels are not equally spaced. By having the less-probable higher levels farther apart, the increased quantization error that results is more likely to be masked by the signal, and, because it’s less probable, the average signal-to-noise level does not increase. In fact, for a given number of levels, the average signal to noise reduces because the more likely levels are closer together and thus generate less quantization noise. MPEG-1 layer 3 uses a nonuniform quantizer to quantize the filtered signal samples. — David M. Howard, Jamie Angus, Acoustics and Psychoacoustics, 2664

Categorization

• SHARE: A niche-but-neat coding trick: exploit masking probability to spend quantization error where the ear won’t notice it.
• DEFINE: Defines nonuniform/nonlinear quantization and explains why it beats uniform quantization.
• corroborated: 3 independent passages support this claim — David Howard, Jamie Angus, coustics and Psychoacoustics, 2662; David M. Howard, Jamie Angus, Acoustics and Psychoacoustics, 2664