This script listens to a small set of voice commands in US English and translates them into instructions for a MythTV frontend. The intention is to emulate the voice control capabilities of an Xbox360 with a Kinect (e.g., you can say "Xbox pause" and it will cause the currently playing video to pause).
The script is fully functional; however, I don't have a good handle on providing echo cancellation, so a) the script will pick up audio from the playing television show and occasionally consider it to be a valid command, and b) it's almost impossible to issue voice commands while a show is playing or when there's other substantial background noise. Suggestions and pull requests would be most welcome. I'm testing the webrtc echo cancelling functionality Pulseaudio 2.0 now, so I'll update here if I can figure something out.
Note: I've been using the voice trigger "TiVo"; this is my own nickname for my MythTV HTPC, and used without permission of the TiVo corporation.
- Linux
- Pocketsphinx 0.5+
- Gstreamer 0.10+
- MythTV 0.24+
- Ubuntu 11.10, 12.04
- MythTV 0.24, 0.25
- This code started as the demo code from Using PocketSphinx with GStreamer and Python.
- Makes use of the MythTV Frontend control socket and MythTV Python bindings.
- Also useful: mythremctl.py.
- Language model and dictionary generated using the CMUSphinx online tool lmtool.
- Using the file
sentence_corpus.txtfrom this repo. - Other sphinx language model resources.
The original link for the LICENSE file of the demo code I used is 404, but the Internet Archive has a copy. This project adheres to that license.