This is not necessarily Pi specific - but will affect anyone (including me!) who has been using this set of instructions to perform speech to text functions on the Pi.
It appears that the v1 version of the Google Speech API used in this has been deprecated.
I had a beautifully running app on the Pi - until last night. There is a v2 version of the api but I have been unable to get it to work with the Python script I have running on the Pi.
My original POST request was....
wget -q -U "Mozilla/5.0" --post-file file.flac --header "Content-Type: audio/x-flac; rate=44100" -O - "http://www.google.com/speech-api/v1/recognize?lang=en-us&client=chromium" | cut -d\" -f12 >stt.txt
Which worked every time with a word perfect response saved to stt.txt - until last night. Grrr.
Have now updated this to....
wget -q -U "Mozilla/5.0" --post-file file.flac --header "Content-Type: audio/x-flac; rate=44100" -O - "http://www.google.com/speech-api/v2/recognize?lang=en-us&client=chromium&key=AIzaSyCnl6MRydhw_5fLXIdASxkLJzcJh5iX0M4" | cut -d\" -f12 >stt.txt
Details of v2 api and details of the 'key' element, I found here... https://github.com/gillesdemey/google-s ... /README.md
I can now connect to server, and get a reponse.I am writing in Python & the original v1 api call returned a perfect response everytime and was saved in stt.txt which I am then passing to another part of the app.
I now get a 200 OK response from google and I am receiving an approx 400b file (which seems tiny to me) back from the v2 api and this is written to SDTOUT.
But - the response that now gets saved to stt.txt is either the single word 'final' or I have twice had the word 'transcript' - and I have no idea why or what this means.
In one attempt with the identical settings, I got a portion of the sentance that I had spoken - every other test has produced the result above.
If anyone has got futher than me - all help is greatfully received. I had a fully working app running on the Pi until last night.