I’m working on the integration of kiku (voice recognition to control your OS) to Vinux (Linux for the visually impaired). For now the feedback about kiku is great, but one thing people wants is a solution for dictation (that is, speech to text). kiku could be use for that task, but sadly there is no good (accurate) acoustic model available under GPL (Voxforge is a good start, but not there yet – please contribute).
So i came up with this online speech recognition for dictation solution to overcome this limitation. Speak in different languages, spell check, translate and text to speech all in one place. You’ll need Google Chrome and a microphone.