pocketsphinx

Great question. This is technically referred to as "Wake Word Detection". You run a really small model locally that is just processing 500ms (for example) of audio at a time through a light weight CNN or RNN. The idea here is that it's just binary classification (vs actual speech recognition). There are some open source libraries that make this relatively easy: - https://github.com/Kitt-AI/snowboy (looks to be shutdown now) - https://github.com/cmusphinx/pocketsphinx This avoids having to stream audio 24x7 to a cloud model which would be super expensive. This being said, I'm pretty sure what the Alexa does, for example, is send any positive wake word to a cloud model (that is bigger and more accurate) to verify the prediction of the local wake word detection model AFAIK. Once you are positive you have a positive wake word detected - that'

It can use https://github.com/cmusphinx/pocketsphinx (offline) or Kaldi(offline), or Google(online) for the "speech-to-text"-part.

Kaldi has a deep learning speech-to-text model now: http://kaldi-asr.org/models.html Or you could use Sphinx https://github.com/cmusphinx/pocketsphinx

You need to try CMUSphinx http://cmusphinx.sourceforge.net. In keyword spotting mode it is very lightweight. Python example is here: https://github.com/cmusphinx/pocketsphinx/blob/master/swig/p...

Haven't tried it yet, but I hear Pocketsphinx has a keyword spotting mode. https://github.com/cmusphinx/pocketsphinx

PocketSphinx[1] isn't bad. I'm sure one could get better accuracy with a DNN recognizer, but good luck getting something like Kaldi to run on a smartphone. 1: https://github.com/cmusphinx/pocketsphinx 2: http://kaldi-asr.org/

> What verifiably private options are there for speech-to-text? Unfortunately not a lot. That's why I said it was "sort of" and answer. I've tried http://julius.osdn.jp/en_index.php and https://github.com/cmusphinx/pocketsphinx in the past, with very limited results. Though it's been a while and they may have gotten better.