Voicely is a voice command interface for Linux, and a fork of the Voximp project. Voicely uses Gstreamer and CMU Sphinx to provide speech recognition, and acts simply as a framework for triggering scripts or other code execution with a keyword or keyphrase spoken to an active audio input device.

The main change which this project makes to the original Voximp project is the reworking of the configuration file from a .py into a .conf with a more sane syntax.

Voicely is designed to be easy to configure, and to offer as much flexibility as possible to users in terms of what commands can do. At present, Voicely supports only one execution mode: shell. By editing voicely.conf, users can create their own commands which when recognized will be executed on a sub-process shell.

Future versions of voicely will also support embedded python in the config file (which will be executed via the eval() function so take care) and commands which require user verification before they execute.

# Sample config file entry
[MUTE ALL AUDIO]
cmd = for i in $(pacmd list-sinks | grep index | sed s/"[index \*:]"/""/g); do; pacmd set-sink-mute $i 1; done;
msg = "Muting all audio devices"
confirm = no 
# truth would present a warning dialog before cmd   [OPTIONAL] [IN DEV]
python  = no 
# truth will eval() cmd within voicely              [OPTIONAL, DEFAULTS TO FALSE] [IN DEV]

Installing Voicely is easy, just run the included install.sh

If you want to do the installation by hand you must

- create the directory ~/.config/voicely/
- move the contents of config/ to the newly created directory
- move voicely.py to /usr/bin/voicely
- chmod it +x
- know you have all the dependencies satisfied

* python2
* python-gobject
* pygtk
* python-wnck
* gstreamer0.10-python
* xdotool
* sphinxbase
* pocketsphinx

Due to limitations of voice recognition, for Voicely or even CMU Sphinx to work you will probably need to keep your microphone at maximum volume at all times for the computer to "hear" you.

Because of this, the computer will also "hear" everything you say and your command phrases should be chosen to prevent accidental action on the part of the computer. One tool which Voicely provides to assist in this is the keyphrase noted in voicely.conf. As its name suggests, the keyphrase emulates keying a microphone to speak to the computer by auto-magically prefixing all commands with the keyphrase.

Say for instance one command is "Google". The odds that you say the word Google aloud in a way which you did not intend Voicely to interpret is probably high. However by adding a keyphrase to the config file and setting the "use keyphrase" flag you would have to say for instance... "Hal google" to achieve the same effect.

Also due to limitations of speech recognition technology, the response time of recognizing a command while under half a second is noticeable. This means that in most cases for my fellow productivity nuts keyboard hot keys and a good desktop will offer better reaction times than Voicely. However as evidenced by my example command MUTE ALL AUDIO, for some more involved tasks which could be achieved via complicated commands, scripts or GUI voice command proves a viable wrapper on top of shell scripting.