audio2numpy load an audio file and directly ouputs the audio data as a numpy array and its sampling rate. Supports .wav, .aiff via python's standard library, and .mp3 via ffmpeg.
Using pip:
pip install audio2numpy
audio2numpy requires ffmpeg to decode mp3 files. You would need to install ffmpeg in order to have mp3 support.
homebrew install ffmpeg
sudo apt-get install ffmpeg
Check here for other installation methods for different Linux distributions.
- Download the latest distribution from https://ffmpeg.zeranoe.com/builds/
- Unzip the folder, preferably to
C:\ - Append the FFmpeg binary folder (e.g.
C:\ffmpeg\bin) to the PATH system variable (How do I set or change the PATH system variable?)
!apt install ffmpeg
from audio2numpy import open_audio
fp = "./examples/word.mp3" # change to the correct path to your file accordingly
signal, sampling_rate = open_audio(fp)
0.1.3 (17.10.2019)
- Add Colab ffmpeg installation guide.
- Remove ffmpeg from requirements.txt. Users should install it separately
0.1.2 (20.08.2019)
Add instructions to install ffmpeg if load mp3 failed with ffmpeg backend not available.
0.1.1 (14.08.2019)
Initial release.