I'm trying to find a method to read an audio file, detect the onset of speech and the offset and return the precise time (in milliseconds) of both.
Are there any good packages for python that can do this, and if so can you also direct me to sample code?
Is there software like like praat, but more precise?
Google cloud wavenet / speech to text is not precise enough.
I tried to do this with librosa but I didn't figure out how to just mark the voice onset and offset. Instead librosa marks the onset of every note in the file, rather than just the beginning and end of speech.
Suggested packages?