shareresearchteam / ReminiscenceStudy

Reminiscence Study Code

Based on https://blog.duy-huynh.com/build-your-own-voice-assistant-and-run-it-locally/

To setup:

• install Ollama (https://ollama.com/)
  • It's just an executable
• run ollama pull gemma:2b-instruct-v1.1-q4_K_M
  • Replace gemma:2b-instruct-v1.1-q4_K_M with any other model from Ollama if desired, just requires a one line change in app.py where the model is setup
• run ollama serve
• run pip install -r requirements.txt
  • linux may complain about pyaudio when running this so run the following and then the requirements line
  • sudo apt-get install portaudio19-dev
  • pip install pyaudio

To run:

• in text only mode for debugging python app.py -1
• in no_rails mode which only lets researcher confirm or regnerate python app.py 1
• in three prompts mode which generates 3 prompts for the researcher to choose from python app.py 2
• in safe mode which has better? prompt engineering python app.py 3
• NOTES:
  • Linux may throw an error for libespeak, run sudo apt install espeak for getting tts to work
  • Linux may throw an error for ffmpeg, run sudo apt install ffmpeg

On WINDOWS:

• tts.setProperty('voice', voices[1].id) ON LINUX
• tts.setProperty('voice', 'english')