Project description

This is voice assistant prototype.

Frontend html + css + js

Backend flask + redis

Supported features

Voice or text input. Speech to text conversion is done with Google Cloud Speech API
Translation from Russian to English using Google Translate API
Translation history (can be removed)
Voice output. Speech synthesis is implemented with gTTS

Usage

For performing queries to Google Speech API the key needed. Possible options:
- Set up authentication by yourself (see tutorial for details). It's free.
- Contact me and I'll share my key (unfortunately I cannot share the key publicly)
To compose docker image:

docker build -t voice-assistant .
docker-compose up

Due to security reasons audio recording with navigator.mediaDevices.getUserMedia cannot be performed on non-security connections (including localhost). To resolve issue, generate SSL for localhost. Then after docker-compose run

localhost:5000

I have not found any other way for audio to be recorded or played on localhost (if you run 0.0.0.0:5000 audio will not be recorded and played).

Sometimes Google Translate API doesn't recognize queries for unknown reasons. You'll get 'Not recognized' output. Important: this is not application error, this is error response from Google's API.