blacklight / Takk

Speech recognition in Python made easy and flexible

Takk

** IMPORTANT: TAKK IS PART OF THE ARMANDO PLATFORM https://github.com/BlackLight/Armando **

** THEREFORE USE IT BY CLONING THE ARMANDO PLATFORM GIT PROJECT **

Takk is more or less a Python and Unix-friendly version of Google Now, Siri or Cortana. It allows you to create custom speech-based commands on your Unix box, relying on the Google Speech Recognition API.

Requirements

• flac executable (check your OS distribution on how to install it - e.g. apt-get install flac on Debian/Ubuntu or pacman -S flac on Arch Linux), in order to convert raw WAV recording in the FLAC format used by Google Speech Recognition API.

• pyaudio (e.g. pip install pyaudio).

• requests module (e.g. pip install requests), used to make HTTP requests to the Google Speech backend.

• phue in case you want to use Takk together with Philips Hue lightbulbs.

Installation

1. git clone https://github.com/BlackLight/Armando.git
2. cd Armando
3. git submodule init
4. git submodule update
5. cd share/Takk
6. Modify ./takkrc to include your Google Speech Recognition API secret key. Instructions on how to get one: http://www.chromium.org/developers/how-tos/api-keys
7. Customize ./takkrc according to the provided in-file documentation
8. ./takk.py