Librosa - How to create mel-spectrogram for stereophonic audio?

Question

There is a code which generates mel-spectrogram for mono audio.

import librosa, librosa.display
import matplotlib.pyplot as plt
import numpy as np

file = "C:/Users/User/Desktop/sound.wav"

y, sr = librosa.load(file, sr=48000)

librosa.feature.melspectrogram(y=y, sr=sr)

S = librosa.feature.melspectrogram(y=y, sr=sr, n_mels=128, fmax=12000)
S_dB = librosa.power_to_db(S, ref=np.max)
lr.display.specshow(S_dB, x_axis='time', y_axis='mel', sr=sr, fmax=12000)

plt.colorbar(format='%+2.0f dB')
plt.title('Mel-spektrogram')
plt.tight_layout()
plt.show()

Is there any possibility to rewrite the code so that it can generate mel-spectrogram for stereophonic audio? If not how else can i do it?

Answer 1

To get multi-channel data you need to use librosa.load(..., mono=False). The shape of y will then be (channels, samples). You can then compute the mel-spectrogram separately for each of the channels.