使用python从给定的音频文件中检测静音索引

Question

I am trying to process an audio file in python using various modules like numpy, struct etc. But I am really having a hard time detecting silence in the file, as in where is the presence of silence. one on the methods I came across was to slide a window of fixed time interval over my audio signal and record the sum of squared elements. I am new to python and hardly aware of it thus unable to implement this method.

Answer 1

If you are open to outside libraries, one of the quick way to do is using pydub .
pydub has a module called silence that has methods detect_silence and detect_nonsilent that may be useful in your case.
However, the only caviar is that silence needs to be at-least half a second.

Below is a sample implementation that I tried using an audio file. However, since silence in my case was less than half a second, only few of the silent ranges were correct.

You may want to try this and see if it works for you by tweaking min_silence_len and silence_thresh

Program

from pydub import AudioSegment,silence


myaudio = intro = AudioSegment.from_wav("a-z-vowels.wav")

silence = silence.detect_silence(myaudio, min_silence_len=1000, silence_thresh=-16)

silence = [((start/1000),(stop/1000)) for start,stop in silence] #convert to sec
print silence

Result

Python 2.7.9 (default, Dec 10 2014, 12:24:55) [MSC v.1500 32 bit (Intel)] on win32 Type "copyright", "credits" or "license()" for more information.

================================ RESTART ================================

[(0, 1), (1, 14), (14, 20), (19, 26), (26, 27), (28, 30), (29, 32), (32, 34), (33, 37), (37, 41), (42, 46), (46, 47), (48, 52)]

Answer 2

For better result use dBFS

from pydub import AudioSegment,silence

myaudio = intro = AudioSegment.from_mp3("RelativityOverview.mp3")
dBFS=myaudio.dBFS
silence = silence.detect_silence(myaudio, min_silence_len=1000, silence_thresh=dBFS-16)

silence = [((start/1000),(stop/1000)) for start,stop in silence] #in sec
print(silence)