Description
Hallo,
would it be possible for the listen method of the Recognizer class to accept one more callback that will be called when Snowboy has detected the keyword but the phrase is not yet recorded. Looking at the code, this is currently not possible but it would be easily implementable. My assumption is that the callback should be called somewhere around line 674 of the init.py file.
A possible problem with this new callback: if the called function takes too much time, this could mess up the recording of the audio in the buffer. I don't understand the actual recording of the audio in the buffer well enough to know, if some delay between line 675 and line 676 in the init.py file would be a problem.
I would be happy to implement the feature myself, if you find the request reasonable.