VoxShell

VoxShell is a simple command and control dialog manager that uses the Julius speech recognition engine and VoxForge acoustic models.

In the most generic sense, the term 'shell' means any program that a user employs to type commands. Unlike a regular shell, VoxShell allows you to speak key words or phrases into a microphone to execute those commands. You can set up grammars to execute any shell command or program using your own voice.

Julius is a high-performance, two-pass large vocabulary continuous speech recognition (LVCSR) decoder. VoxForge is a 'Free' speech repository (speech corpus) that uses the collected speech to create acoustic models.

Try it out

follow instructions in INSTALL.md.

Linux

run VoxShell version of julius

  $ bin/voxshell -input mic -C share/voxshell/en/voxshell.jconf -gramlist share/voxshell/en/grammar_list

how do grammar files work?

Look to the .vox and .grammar grammar files to see what the system will recognize (e.g. language/en/linux/rhythmbox/rhythmbox.vox & rhythmbox.grammar).

If you modify one of the grammar files, just run make grammar in the top voxshell directory to recompile your all your grammar files (i.e. creates new .dict, .dfa, .term files).

Windows:

MingW or Cygwin console:

  $ bin/voxshell -input mic -C share/voxshell/en/voxshell.jconf -gramlist share/voxshell/en/grammar_list

Command Prompt window:

  C:> bin\voxshell.exe -input mic -C share\voxshell\en\voxshell.jconf -gramlist share\voxshell\en\grammar_list

Submit speech samples to make better acoustic models

We need many more speech submissions to create high quality free Acoustic Models. Please take the time to submit some speech to www.voxforge.org using the VoxForge speech applet.

thank you!

VoxForge/voxshell