python - Librosa:过滤奈奎斯特以外的通带

Question

Librosa Python 3.5 在 8000Hhz 音频样本上出现错误:

extract feature error. Filter pass-band lies beyond Nyquist

之前遇到过有关 Nyquist 的其他错误，但通过手动设置 sr 和 fmin 得到修复，但似乎无法通过此错误。

stft = np.abs(librosa.stft(X))
mfccs = np.mean(librosa.feature.mfcc(y=X, sr=8000, n_mfcc=40, fmin=60).T,axis=0)
chroma = np.mean(librosa.feature.chroma_stft(S=stft, sr=8000).T,axis=0)
mel = np.mean(librosa.feature.melspectrogram(X, sr=8000, fmin=60).T,axis=0)
contrast = np.mean(librosa.feature.spectral_contrast(S=stft, sr=8000, fmin=60).T,axis=0)
tonnetz = np.mean(librosa.feature.tonnetz(y=librosa.effects.harmonic(X), sr=8000).T,axis=0)
return mfccs, chroma, mel, contrast, tonnetz

编辑:

tonnetz 功能似乎存在问题。每docs色度:

np.ndarray [shape=(n_chroma, t)] or None

    Normalized energy for each chroma bin at each frame.

    If None, a cqt chromagram is performed.

cqt 色度图需要额外的参数才能与 8000Hhz 音频一起使用，因此我需要手动设置色度。不幸的是，这超出了我的 Librosa/Audio 知识范围，有人为我提供了线索吗？

Answer 1

    stft = np.abs(librosa.stft(X))
    mfccs = np.mean(librosa.feature.mfcc(y=X, sr=8000, n_mfcc=40, fmin=30).T,axis=0)
    chroma = np.mean(librosa.feature.chroma_stft(S=stft, sr=8000).T,axis=0)
    mel = np.mean(librosa.feature.melspectrogram(X, sr=8000, fmin=30).T,axis=0)
    contrast = np.mean(librosa.feature.spectral_contrast(S=stft, sr=8000, fmin=30).T,axis=0)
    tonnetz = np.mean(librosa.feature.tonnetz(y=librosa.effects.harmonic(X), sr=sample_rate, chroma=librosa.feature.chroma_cqt(y=X, sr=8000, fmin=30)).T,axis=0)
    return mfccs, chroma, mel, contrast, tonnetz