tyiannak / segment_sound

A Python script to break an audio signal into audio segments using silence removal

segment_sound: A Python script to break an audio signal into audio segments using silence removal

segment_sound is a basic script to split a WAV file into segments based on the existence of pauses between the corresponding segments. It uses the silence removal function silence_removal() frome the pyAudioAnalysis library.

Requirements:

Tested in > Python3.6. To install requirements: pip install -r requirements.txt

Usage:

Execution:

python3 segment_audio.py -i <path_to_wav_file> -s <smoothing_window_in_secs> -t <energy_thres_0_to_1> -o <path_to_out_folder>

E.g

python3 segment_audio.py -i obama_speech_1.wav -s 0.75 -t 0.32 -o obama_segments_1

The above example results in the following:

Output audio segments are stored in WAV files in the output folder (provided as arg):

 ⇒ ls -1 obama_segments_1 
obama_speech_1_0.18_2.36.wav
obama_speech_1_11.32_13.44.wav
obama_speech_1_4.0200000000000005_5.2.wav
obama_speech_1_6.5600000000000005_9.9.wav

You obviously need to experiment with the two parameters depending on the level of background noise and on rate of the audio events (e.g. faster speech may require smaller smoothing window).