mic > keyword spotting (pocketsphinx) > tts (bing) > text parser > [stt >] > speaker
- Clone this repo and get pocketsphinx acoustic model, place the files as the following structure.
cd /tmp/run/mountd/mmcblk0p1 # suppose we place files to a sd card
git clone https://github.com/respeaker/respeaker_hi.git
git clone https://github.com/respeaker/pocketsphinx_keyword_spotting.git
cp -R pocketsphinx_keyword_spotting/model/hmm respeaker_hi/model
cd respeaker_hi
- Get a key from # get a key from https://www.microsoft.com/cognitive-services/en-us/speech-api and create creds.py with the key
cp creds_template.py creds.py
vi creds.py # add the key and save
- Run
python main.pyto start a journey
The files structure will be like:
respeaker_hi
│ bing_recognizer.py
│ creds_template.py
│ creds.py
│ main.py
│ microphone.py
│ player.py
│ README.md
├─audio
│ hi.wav
└─model
│ respeaker.dic
│ keywords.txt
└─hmm
└─en
feat.params
mdef
means
noisedict
README
sendump
transition_matrices
variances