dspeech.feature
Members
Functions
- dctauto dct(S xs)
Computes type-II DCT on 1-d signal:
- ditherauto dither(S wave, double stddev = 1.0)
Dither quantized wave
- melMatrixmelMatrix(size_t nFreq, size_t nMel, Dtype sampleFreq, Dtype lowFreq = 20, Dtype highFreq = 0)
Computes Mel-scale filterbank matrix
- preemphasisauto preemphasis(S wave, double factor = 0.97)
Preemphasis wave signal
- spectrogramauto spectrogram(S xs, size_t nperseg = 256, size_t noverlap = 128)
Computes a magnitude 2-D spectrogram from a time-domain 1-D signal
- toFbankauto toFbank(S spect, M melmat)
Computes log-Mel filterbank (FBANK) feature from spectrogram signal and Mel matrix
- toMfccauto toMfcc(S fbanks, size_t nceps = 13, double lifter = 22)
Computes MFCC feature from log-Mel filterbank (FBANK) feature
Examples
Compute Wave -> Spectrogram -> FBANK -> MFCC eventually
- preemphasised and dithered wave
- power spectrogram
- log-Mel filterbank (FBANK)
- Mel-frequency cepstral coefficients (MFCC)
1 import dspeech.plot : docDir, docWav, plotVector, plotMatrix; 2 import dffmpeg : Audio; 3 import mir.ndslice : map, sliced; 4 5 auto raw = Audio!short().load(docWav); 6 auto wav = raw.data.sliced.dither(1.0).preemphasis(0.97); 7 wav.plotVector.save(docDir ~ "dspeech.feature.speech_wav.png", 400, 200); 8 auto sp = wav.spectrogram(512, 256); 9 sp.plotMatrix.save(docDir ~ "dspeech.feature.speech_spectrogram.png", 400, 200); 10 auto fbanks = sp.toFbank(melMatrix(sp.length!0, 80, cast(double) raw.sample_rate)); 11 fbanks.plotMatrix.save(docDir ~ "dspeech.feature.speech_fbank.png", 400, 200); 12 auto mfccs = fbanks.toMfcc; 13 mfccs.plotMatrix.save(docDir ~ "dspeech.feature.speech_mfcc.png", 400, 200);
Speech feature extraction module
See also: - "Chapter 5 Speech Input/Output," HTK book 3.5 alpha1, pp. 89 - 1 - https://github.com/jameslyons/python_speech_features
TODO: preemphasis and dither