The project aims to use keyword recognition streaming in a real-time offline embedded system using RaspberryPi3.
pip install -U -r requirement.txt- Run with Tensorflow
python tensor_speech_command.py- Run with TensorflowLite
python speech_command.py- The keyword recognizing process has two stages.
- The trigger word is Marvin.
- When you speak Marvin, if the system recognizes the trigger word, the chime sound is played.
- Move to stage 2
- The command keywords are zero, one, two, three, four, five, six, seven, eight, nine, up, down, on, off.
- The system will recognize the command keyword in control_time s.
- If the system recognizes the command keyword, it will execute the corresponding command for that keyword.
- Until the time for this process is greater than control time, it will execute stage 1.