Bug 162844

Summary: continuous voice recognition as action triggers
Product: [Unmaintained] khotkeys Reporter: Manuel Amador (Rudd-O) <rudd-o>
Component: Voice recognitionAssignee: Olivier Goffart <ogoffart>
Status: RESOLVED UNMAINTAINED    
Severity: wishlist CC: ab4bd, nate, toddrme2178
Priority: NOR    
Version First Reported In: unspecified   
Target Milestone: ---   
Platform: Ubuntu   
OS: Linux   
Latest Commit: Version Fixed/Implemented In:
Sentry Crash Report:

Description Manuel Amador (Rudd-O) 2008-05-29 23:41:59 UTC
Version:            (using KDE 3.5.9)
Installed from:    Ubuntu Packages
OS:                Linux

This "press key to start recording + press key to stop recording" monkey business is not good enough.  It would be great if khotkeys had an option to detect utterances at all times, perhaps coupled with an attention word.    That way, I can say:

computer, mute

and the computer would obey.

What technical challenges need to be resolved to get there?
Comment 1 Todd 2008-10-01 18:44:49 UTC
Voice control is a great feature, and is one of the few areas where Linux and OSS still falls way behind windows and closed-source (along with voice recognition in general).

As best as I can tell there are currently 2 open-source speech recognition/voice control projects still in active development.

One is Julius, which is apparently more academically-oriented and more focused on voice recognition than voice control, but may be adaptable to voice control:
http://julius.sourceforge.jp/en_index.php

The other is CMU Sphinx, with the PocketSphinx sub-project appearing to have the most development.  PocketSphinx appears to be focused particularly on voice control, particularly for handheld devices but it should work well for desktop systems as well:
http://cmusphinx.sourceforge.net/
http://www.speech.cs.cmu.edu/

There is also perlbox's voice plugin, which I understand was fairly successful and had a KDE 3.x plugin but does not appear to still be under active development.  It was under development as recently as 2007, though, which is much more recent than most open source voice recognition projects:
http://perlbox.org/

VoxForge is an online open source speech corpus (recorded and transcribed human speech) that is meant to be used with open-source software.  Julius uses it:
http://voxforge.org/

Gnome is currently working on a voice control project based on pocketsphinx:
http://live.gnome.org/GnomeVoiceControl
Comment 2 Nate Graham 2024-09-23 20:56:35 UTC
KHotkeys was removed in Plasma 6; closing its old bug reports.