To compute spectrograms and resample audio wave

tasks/lib/preprocessing.py

@@ -0,0 +1,8 @@
+import librosa
+import numpy as np
+
+
+def resample_audio(raw_wave: np.ndarray, orig_sr: int, target_sr: int) -> np.ndarray:
+    if orig_sr == target_sr:
+        return raw_wave
+    return librosa.resample(raw_wave, orig_sr=orig_sr, target_sr=target_sr)

tasks/lib/spectrogram.py

@@ -0,0 +1,37 @@
+from math import log2
+
+import librosa
+import numpy as np
+
+
+def _get_n_fft(freq_res_hz: int, sr: int) -> int:
+    """
+    :freq_res: frequency resolution in Hz = sample_rate / n_fft
+              how good you can differentiate between frequency components
+              which are at least ‘this’ amount far apart.
+    :sr: sampling_rate
+
+    The n_fft specifies the FFT length, i.e. the number of bins.
+    Low frequencies are more distinguishable when n_fft is higher.
+    For computational reason n_fft is a power of 2 (2, 4, 8, 16, ...)
+    """
+    return 2 ** round(log2(sr / freq_res_hz))
+
+
+def get_spectrogram_dB(
+    raw_wave: np.ndarray, freq_res_hz: int = 5, sr: int = 12000
+) -> np.ndarray:
+    spectrogram_complex = librosa.stft(y=raw_wave, n_fft=_get_n_fft(freq_res_hz, sr))
+    spectrogram_amplitude = np.abs(spectrogram_complex)
+    return librosa.amplitude_to_db(spectrogram_amplitude, ref=np.max)
+
+
+def get_mel_spectrogram_dB(
+    raw_wave: np.ndarray, freq_res_hz: int = 5, sr: int = 12000
+) -> np.ndarray:
+    spectrogram_complex = librosa.stft(y=raw_wave, n_fft=_get_n_fft(freq_res_hz, sr))
+    spectrogram_amplitude = np.abs(spectrogram_complex)
+    mel_scale_sepctrogram = librosa.feature.melspectrogram(
+        S=spectrogram_amplitude, sr=sr
+    )
+    return librosa.amplitude_to_db(mel_scale_sepctrogram, ref=np.max)