Digital Signal Processing Reference
In-Depth Information
Level
80
dB
fm = 250 Hz
fm = 1 kHz
fm = 4 kHz
60
40
20
0
f
0 ,02
0, 05
0, 1
0, 2
0 ,5
1
2
5
10
20 kH z
Level
100
80
dB
60
40
20
0
f
20 kH z
0,02
0,05
0 ,1
0,2
0 ,5
1
2
5
10
Illustration 247:
Masking and threshold of audibility
Top diagram: A loud 250 Hz tone causes a masking area or masks an area so that the quieter tones in the
vicinity cannot be heard.
Bottom diagram: the height and width of the masking area increases considerably with loudness; at the
tolerance limit of 100 decibels it ranges from roughly 200 to 20,000 Hz.
In the case of the state of the art MPEG audio encoding (MPEG = moving pictures expert
group; it is responsible for compression processes for digital audio and video; its
standards are recognised worldwide) for the two reasons given the frequency band of the
audio-signal is divided up into 32 frequency bands of equal size. Each of these frequency
bands contains a narrowband filtered part of the original audio signal. For each of these
bands the masking properties are exploited. Weak tones (frequencies) are eliminated if
loud ones are present, and at the same time the coarseness of the quantisation is adapted
to the masking.
