aspen.stimuli.noise_vocoded_speech¶
Noise-vocoded speech
Classes
|
Generate nosie-vocoded speech. |
-
class
aspen.stimuli.noise_vocoded_speech.NoiseVocodedSpeech(samp_freq=16000, num_freqband=4, freqband_scale_method='octave', user_freqband='0_600_1500_2100_4000', freqband_limit='500_8000', erb_band_number_limit='3_35', erb_band_number_step=1, filter_impulse_response_method='fir', filter_order=512, filter_fir_window='hann', ext_env_method='rect', ext_env_impulse_response_method='fir', ext_env_filter_order=512, ext_env_fir_window='hann', ext_env_freq=16)[source]¶ Bases:
aspen.interfaces.abs_common_interface.AbsCommonInterface,aspen.interfaces.abs_stimulus_interface.AbsStimulusInterfaceGenerate nosie-vocoded speech.
- Parameters
num_freqband¶ (
int) – Number of frequency band. Defaults to 4.freqband_scale_method¶ (
str) – Scale of frequency band. The choices are octave, erb and user. Defaults to “octave”.user_freqband¶ (
str) – User defiened frequency band scale. Use only when freqband_scale_method is user. Defaults to “0_600_1500_2100_4000”.freqband_limit¶ (
str) – Lower & upper frequency of bandpass. Use only when freqband_scale_method is octave or erb. Defaults to “500_8000”.erb_band_number_limit¶ (
str) – Lower & upper bandlimit number in ERB scale. Use only when freqband_scale_method is erb. Defaults to “3_35”.erb_band_number_step¶ (
int) – Step number in ERB scale to reduce the ERB. Use only when freqband_scale_method erb. Defaults to 1.filter_impulse_response_method¶ (
str) – Type of impulse response for filtering. Defaults to “fir”.filter_order¶ (
int) – Number of the filter order. Defaults to 512.filter_fir_window¶ (
str) – Type of FIR window for filtering. Window function is listed in Scipy doc (https://docs.scipy.org/doc/scipy/reference/signal.windows.html). Defaults to “hann”.ext_env_method¶ (
str) – Method of envelope extraction. hilbert is (hilbert transform + low-pass filter). rect is (half-wave rectification + low-pass filter). Defaults to “rect”.ext_env_impulse_response_method¶ (
str) – Type of impulse response for extracting envelope. Defaults to “fir”.ext_env_filter_order¶ (
int) – Number of the filter order for extracting envelope. Defaults to 512.ext_env_fir_window¶ (
str) – Type of FIR window for extracting envelope. Window function is listed in Scipy doc (https://docs.scipy.org/doc/scipy/reference/signal.windows.html).”, Defaults to “hann”.ext_env_freq¶ (
float) – Frequency of the lowpass filter for extracting envelope. Defaults to 16.
-
__call__(x)[source]¶ Generate noise-vocoded speech.
- Parameters
x¶ (
Sequence[ndarray]) – Speech signal and noise signal. x must be sequence-like object such as list, tuple and so on. The first element is speech signal and the second one is noise (i.e. [speech, noise]). The duration of noise must be equal to or greater than the one of speech signal. In the case of gap_method=silent, the second element is ignored.- Return type
- Returns
Noise-vocoded speech.