This operation is needed for the psychoacoustical analysis of the signal, since tonal and non-tonal components give different contribution to setting of the masking threshold.
Tone characteristics depend on the frequency range, since the human ear analyzes the signal split in critical subbands and, as already mentioned, critical bands have an amplitude that vary in function of the frequency. The width of the critical bands takes values from 0.1 kHz for low frequencies up to 4 kHz for high ones.
The TONAL components are set to correspond with the global maxima of the spectrum obtained from the FFT within a certain frequency interval.
The frequency interval within which it has to be verified whether a local maximum is also a global maximum, thus identifying a tonal component, depends on the sampling frequency, on the layer and its spectral location:
| SAMPLING FREQUENCY | LAYER | WIDTH INTERVAL | FREQUENCY |
|---|---|---|---|
| 32 KHZ | I | 125 HZ | 0.0 KHZ < f < 4.0 KHZ |
| 187.5 HZ | 4.0 KHZ < f < 8.0 KHZ | ||
| 375 HZ | 8.0 KHZ < f < 16.0 KHZ | ||
| II | 62.5 HZ | 0.0 KHZ < f < 3.0 KHZ | |
| 93.75 HZ | 3.0 KHz < f < 6.0 KHz | ||
| 187.5 HZ | 6.0 KHz < f < 12.0 KHz | ||
| 375 HZ | 12.0 KHz < f < 16.0 KHz | ||
| 44.1 KHZ | I | 172.266 HZ | 0.000 KHZ < f < 5.512 KHZ |
| 281.25 HZ | 5.512 KHZ < f < 11.024KHZ | ||
| 562.50 HZ | 11.024 KHZ < f < 1KHZ | ||
| II | 86.133 HZ | 0.000 KHZ < f < 2.756 KHZ | |
| 129.199 HZ | 2.756 KHz < f < 5.512 KHz | ||
| 258.398 HZ | 5.512 KHz < f < 11.024 KHz | ||
| 516.797 HZ | 11.024 KHz < f < 19.982 KHz | ||
| 48 KHZ | I | 187.5 HZ | 0.0 KHZ < f < 6.0 KHZ |
| 281.25 HZ | 6.00 KHZ < f < 12.0 KHZ | ||
| 562.50HZ | 12.0 KHZ < f < 24.0 KHZ | ||
| II | 93.750 HZ | 0.0 KHZ < f < 3.0 KHZ | |
| 140.63HZ | 3.0 KHz < f < 6.0 KHz | ||
| 281.25HZ | 6.0 KHz < f < 12.0 KHz | ||
| 562.50 HZ | 12.0 KHz < f < 24.0 KHz |
To identify spectral lines X(k) as tonal or non-tonal, the following three operations are performed.
[ Index | Main Paragraph ]