SpeechToSentiment

This repository integrates Pykaldi Speech Recognition Toolkit with a text classifier to obtain the sentiment of the user's command (locate, describe, no_op).

Steps to setup:

1. Create an environment using Virtualenv or Conda
2. Clone the Pykaldi repository from https://github.com/pykaldi/pykaldi.git and build from source by following its README installation instructions
3. In the zamia directory, run ./models.sh to download the zamia model
4. Set the source to the kaldi installation by running source path.sh
5. Run ./decode_live_smart_test.py to start sentiment analysis