Adaptive loudness levelling method for digital audio signals in frequency domain
Embodiments of the present disclosure are directed to techniques for adjusting the amplitude of a digital audio signal in the frequency domain to control the perceived loudness of the audio signal at a desired level. In one embodiment, a method first adjusts the audio signal to a desired loudness level by applying an adaptive wideband gain and thereafter a multi-band compression is applied to further reduce a dynamic range of the audio signal, and noise analysis and temporal masking operations are also performed to provide a pleasant sound for a listener or listeners.
1. A frequency-domain system for modifying an amplitude of an input digital audio signal, comprising:
a frequency analysis module configured to segment the input digital audio signal into overlapping frames and to apply an analysis window to each frame, and configured to then convert each frame to the frequency domain;
a peak loudness value estimation module configured to estimate a peak loudness value used to derive an adaptive gain for each frame;
an automatic gain control module configured to compute a wideband gain based on the peak loudness values to provide an adjusted digital audio signal in the frequency domain;
a frequency synthesis module configured to convert the adjusted digital audio signal to the time domain;
an overlap-and-add module configured to apply a synthesis window to the adjusted digital audio signal in the time domain to provide an output digital audio signal; and
a multi-band dynamic compressor module configured to separate each frame into signals for sub-bands, each sub-band including at least one frequency bin and the multi-band dynamic compressor module further configured to compute a multi-band gain to compress the signal of each sub-band at a faster speed than the automatic gain control nodule computes the wideband gain, and wherein the multi-band dynamic compressor module is further configured to compute the gain of each bin based on linear interpolation and to then provide linear phase low-pass finite impulse response filtering to generate a gain for each frequency bin.
2. The frequency-domain system of claim 1 further comprising a noise detection module configured to detect background noise based on power information of each of the sub-bands.
3. The frequency-domain system of claim 2 further comprising a silence detection module to detect program switch.
4. The frequency-domain system of claim 1 , wherein the multi-band dynamic compressor module is further configured to smooth the multi-band gain based on noise information and silence information.
5. The frequency-domain system of claim 4 , wherein the multi-band dynamic compressor module includes user configurable parameters that define the operation of the multi-band dynamic compressor module.
6. The frequency-domain system of claim 4 , wherein the multi-band dynamic compressor module is configured to change the gain for each sub-band at a first rate when the sub-band gain is decreasing and at a second rate that is less than the first rate when the sub-band gain is increasing.
7. The frequency-domain system of claim 2 , wherein the noise detection module is further configured to determine whether a current frame is noise or not noise.
8. The frequency-domain system of claim 2 , wherein the noise detection module is further configured to generate an activity score that is further smoothed to avoid excessive fluctuation.
9. The frequency-domain system of claim 8 , wherein the automatic gain control module is further configured to reduce wideband gain fluctuation when the activity score calculated by the noise detection module is below a noise threshold.
10. The frequency-domain system of claim 8 , wherein the multi-band dynamic compressor module is configured to reduce gain fluctuation when the activity score calculated in the noise detection module is below a noise threshold.
11. The frequency-domain system of claim 3 wherein the silence detection module is configured to detect program switch in the input digital audio signal based on an accumulated number of low level frames.
12. The frequency-domain system of claim 11 , wherein the automatic gain control module is configured initialize the wideband gain based on a current gain of a frame when a reset signal is generated by silence detection module.
13. The frequency-domain system of claim 11 , wherein the multi-band dynamic compressor module is configured to initialize the multi-band gain based on a current gain of the frame for each band when a reset signal is generated by the silence detection module.
14. A frequency-domain system for modifying an amplitude of an input digital audio signal, comprising
a frequency analysis module configured to segment the input digital audio signal into overlapping frames and to apply an analysis window to each frame, and configured to then convert each frame to frequency domain;
a peak loudness value estimation module configured to estimate a peak loudness value used to derive an adaptive gain for each frame;
an automatic gain control module configured to compute a wideband gain based on the peak loudness values to provide an adjusted digital audio signal in the frequency domain;
a frequency synthesis module configured to convert the adjusted digital audio signal to the time domain;
an overlap-and-add module configured to apply a synthesis window to the adjusted digital audio signal in the time domain to provide an output digital audio signal; and
a multi-band dynamic compressor module configured to separate each frame into signals for sub-bands, each sub-band including at least one frequency bin and the multi-band dynamic compressor module further configured to compute a multi-band gain to compress the signal of each sub-band at a faster speed than the automatic gain control module computes the wideband gain, and wherein the multi-band dynamic compressor is further configured to compute gains for each sub-band based on a compression curve and a single target level for several sub-bands and wherein the target level in the compression curve is calculated by subtracting a target level associated with the automatic gain control module by a constant.
15. The frequency-domain system of claim 14 , wherein the automatic gain control module calculates the wideband gain based on a four-segment compression curve.
16. The frequency-domain system of claim 14 , wherein the automatic gain control module is configured to change the wideband gain faster when the wideband gain is decreasing and to change the wideband gain slower when the wideband gain is increasing.
17. The frequency-domain system of claim 14 , wherein the multi-band dynamic compressor module is further configured to compute a power for each of the sub-bands.
18. The frequency-domain system of claim 15 , wherein the automatic gain control module is configured to scale the wideband gain based on a slope of the compression curve.
19. The frequency-domain system of claim 17 , wherein the multi-band dynamic compressor module is configured to compute gains for each of the sub-bands based on a four-segments compression curve and a single target level for several sub-bands.
20. A frequency-domain system for modifying amplitudes of a plurality of input digital audio signals, comprising:
a frequency analysis module configured to segment each input digital audio signal into overlapping frames and to apply an analysis window to each frame, and configured to then convert each frame to frequency domain;
a peak loudness value estimation module configured to estimate a peak loudness value used to derive an adaptive gain for each frame;
an automatic gain control module configured to compute a wideband gain based on the peak loudness values to provide for each input digital audio signal an adjusted digital audio signal in the frequency domain;
a frequency synthesis module configured to convert each adjusted digital audio signal to the time domain;
an overlap-and-add module configured to apply a synthesis window to each adjusted digital audio signal in the time domain to provide a corresponding output digital audio signal for each of the plurality of input digital audio signals;
a multi-band dynamic compressor module configured to separate each frame into signals for sub-bands, each sub-band including at least one frequency bin and the multi-band dynamic compressor module further configured to compute a multi-band gain to compress the signal of each sub-band at a faster speed than the automatic gain control module computes the wideband gain; and
a multi-processing channel module configured to generate a combined channel based on a plurality of input channels, each input channel being associated with a corresponding one of the plurality of input digital audio signals, and further configured to generate a gain based on all the plurality of input channels and applied to all the plurality of input channels.
21. The frequency-domain system of claim 20 , wherein the frequency analysis module is further configured to apply to each frame converted to the frequency domain a frequency weighting curve based on ITU-R BS.1770 recommendation to generate frequency weighted data for each frame.
22. The frequency-domain system of claim 21 , wherein the peak loudness value estimation module is further configured to smooth the estimated peak loudness value after the frequency analysis module applies the frequency weighting curve to thereby generate the peak loudness value.
23. The frequency-domain system of claim 22 , wherein the automatic gain control module is configured to generate the wideband gain based on the peak loudness value.
24. The frequency-domain system of claim 22 , wherein the automatic gain control module is further configured to simulates the temporal masking effect of human hearing based on the peak loudness value.
25. The frequency-domain system of claim 22 , wherein the automatic gain control module is configured to reduce wideband gain fluctuation when peak loudness value is below a noise level threshold.
26. The frequency-domain system of claim 22 , wherein the automatic gain control module is further configured to compute a rate of change of the peak loudness value and is configured to increase the change of the wideband gain based on the rate of change of the peak loudness value.
27. The frequency-domain system of claim 22 , wherein the multi-band dynamic compressor module is configured to reduce multi-band gain fluctuation when the peak loudness value is below a noise level threshold.
28. The frequency-domain system of claim 22 , wherein the multi-band dynamic compressor module is configured to simulate the temporal masking effect of human hearing to smooth the multi-band gains for the sub-bands based on the peak loudness of the signal.