TL;DR: I'm using WhisperIMEplus on my phone, and I think I will finally live in the XXIst century with my phone.
https://github.com/woheller69/whisperIMEplus

I refrained myself from using speech recognition on my android since the beginning because I didn't like the idea of my voice used for other reasons than my need which would have been speech-to-text.

And having on-device speech recognition was pretty niche for a while (I was interested in mycroft and snips at that time). Then there was Mozilla with CommonVoice and deepspeech, unfortunately, DeepSpeech has been shut down (it seems), and the results are far from the Whisper model from OpenAI.

I'm clearly not an OpenAI fan (if you haven't figured it out yet, you will soon if you follow me), but Whisper seems to be the best thing that got out of this, mostly because it's way more open than any other things from OpenAI which are not open at all.

Anyway, I found that now, there is a project called WhisperIMEplus that is used as a keyboard on my android, and it processes my voice, locally, on my device. And the app has NO internet connection rights, so, even if OpenAI added some backdoor to send data online in their Whisper model, well, Android app rights wouldn't allow it.

I'm fine with all of this, so now, I can finally take notes by talking to my phone, in English and French, without having second thoughts about it.

It's good when technology helps you, instead of trying to screw you in different and sneaky ways.

#SpeechRecognition #android #privacy #SpeechToText #OpenAI #whisper #ai