Speaker recognition streaming

[thestumonkey]

Currently the speaker service works on a file only. This means that in order to only trigger on my voice we need to convert the stream to a file

[linear]

CHR-58