drc#

diffsptk.DRC#: alias of DynamicRangeCompression

class diffsptk.DynamicRangeCompression(threshold, ratio, attack_time, release_time, sample_rate, makeup_gain=0, abs_max=1, learnable=False)[source]#

See this page for details.

Parameters:

thresholdfloat <= 0: Threshold in dB.
ratiofloat > 1: Input/output ratio.
attack_timefloat > 0: Attack time in msec.
release_timefloat > 0: Release time in msec.
sample_rateint >= 1: Sample rate in Hz.
makeup_gainfloat >= 0: Make-up gain in dB.
abs_maxfloat > 0: Absolute maximum value of input.
learnablebool: Whether to make the DRC parameters learnable.

References

[1]

C.-Y. Yu et al., “Differentiable all-pole filters for time-varying audio systems,” Proceedings of DAFx, 2024.

forward(x)[source]#

Perform dynamic range compression.

Parameters:

xTensor [shape=(…, T)]: Input signal.

Returns:

outTensor [shape=(…, T)]: Compressed signal.

Examples

>>> x = torch.randn(16000)
>>> x.abs().max()
tensor(4.2224)
>>> drc = diffsptk.DynamicRangeCompression(-20, 4, 10, 100, 16000)
>>> y = drc(x)
>>> y.abs().max()
tensor(2.5779)

diffsptk.functional.drc(x, threshold, ratio, attack_time, release_time, sample_rate, makeup_gain=0, abs_max=1)[source]#

Apply dynamic range compression.

Parameters:

xTensor [shape=(…, T)]: Input signal.
thresholdfloat <= 0: Threshold in dB.
ratiofloat > 1: Input/output ratio.
attack_timefloat > 0: Attack time in msec.
release_timefloat > 0: Release time in msec.
sample_rateint >= 1: Sample rate in Hz.
makeup_gainfloat >= 0: Make-up gain in dB.
abs_maxfloat > 0: Absolute maximum value of input.

Returns:

outTensor [shape=(…, T)]: Compressed signal.