Python Audio Libraries

---

• Author: Diego Inácio
• GitHub: github.com/diegoinacio
• Notebook: audio_libraries.ipynb

---

Brief overview about some of the main python libraries which promote input and output of digital audio files.

%matplotlib inline
import matplotlib.pyplot as plt
from IPython.display import Audio
import numpy as np

from _utils import *

Scipy

---

Reading and writing audio files using SciPy.

Installation

---

Install by command line:

>>> conda install scipy

or

>>> pip install scipy

Input

---

from scipy.io import wavfile

%%time
try:
    fs, x_scipy = wavfile.read('../_data/A4_cello01.wav')
    x_scipy = x_scipy.astype(np.float32)
    x_scipy /= np.abs(x_scipy).max(axis=0)
except:
    print('Can not load audio!')

Wall time: 9.97 ms

audiovis(x_scipy, fs)

audio stereo

Audio(x_scipy.min(axis=1), rate=fs)

Analysis

---

summary_audio(x_scipy)

                          L               R
       min:         -0.9645         -0.9377
  1st Quar:         -0.2134         -0.1654
    median:         -0.0060         -0.0115
      mean:         -0.0017         -0.0016
  3rd Quar:          0.1955          0.1379
       max:          1.0000          1.0000
     sigma:          0.3149          0.2663

%%time
spectrogram(x_scipy, fs=fs, ws=2048)

audio stereo

Wall time: 871 ms

Output

---

#wavfile.write('_output/audio_scipy.wav', fs, x_scipy)

pySoundFile

---

Reading and writing audio files using pySoundFile.

Installation

---

Install by command line:

>>> conda install -c conda-forge pysoundfile

or

>>> pip install pysoundfile

Input

---

import soundfile

%%time
try:
    x_soundfile, fs = soundfile.read('../_data/A4_cello01.wav')
except:
    print('Can not load audio!')

Wall time: 5.98 ms

audiovis(x_soundfile, fs)

audio stereo

Audio(x_soundfile.min(axis=1), rate=fs)

Analysis

---

summary_audio(x_soundfile)

                          L               R
       min:         -0.2648         -0.2215
  1st Quar:         -0.0586         -0.0391
    median:         -0.0016         -0.0027
      mean:         -0.0005         -0.0004
  3rd Quar:          0.0537          0.0326
       max:          0.2745          0.2362
     sigma:          0.0865          0.0629

%%time
spectrogram(x_soundfile)

audio stereo

Wall time: 842 ms

Output

---

#soundfile.write('_output/audio_soundfile.wav', x_soundfile, fs)

LibROSA

---

Reading and writing audio files using LibROSA.

Installation

---

Install by command line:

>>> conda install -c conda-forge librosa

or

>>> pip install librosa

Input

---

import librosa

%%time
try:
    x_librosa, fs = librosa.load('../_data/A4_cello01.wav', mono=False)
    x_librosa = x_librosa.T
except:
    print('Can not load audio!')

Wall time: 495 ms

audiovis(x_librosa, fs)

audio stereo

Audio(x_librosa.min(axis=1), rate=fs)

Analysis

---

summary_audio(x_librosa)

                          L               R
       min:         -0.2633         -0.2182
  1st Quar:         -0.0585         -0.0390
    median:         -0.0017         -0.0028
      mean:         -0.0005         -0.0004
  3rd Quar:          0.0536          0.0326
       max:          0.2729          0.2319
     sigma:          0.0865          0.0629

%%time
spectrogram(x_librosa)

audio stereo

Wall time: 643 ms

Saída

---

#librosa.output.write_wav('_output/audio_librosa.wav', x_librosa, fs)