speech -->into--> text

**DJonsson2008** · September 16th, 2010

There are a few threads in this forum including the terms
'speech recognition' or "speech-to-text" or 'text-to-speech'.

Its confusing sorting through all this when what one is interested
in is simply speech-to-text or speech-to-text editor.

Speech recognition means different things to different people.

1. Tell a computer and/or robot what to do with a finite set of commands.
_This includes desktop controls and instructions to run programs.

2. Play text back through an audio speech generator.

3. Speak into a computer and have it turn your voice into text that can be saved in a word processor,
_Of course with enough positive word recognition to save significant keyboarding.

So number 3 is what I'm looking for specifically: speech -->into--> text

Is is possible with Ubuntu/linux to

* Open a word processor and/or text editor
* Turn on the microphone
* And have the spoken words be converted to text

I've read good reviews of SpeakingNaturally for Windows OS, and some
here have suggested that software with Wine.

But I'm wondering if any app or combination of apps
makes this possible with Ubuntu/Linux?

**DJonsson2008** · September 19th, 2010

This is what I've found. Speech into text (meaning full dictation) is apparently not fully developed in Open Source Linux.

The best gui and most fully implemented application for Ubuntu
I can find in this area is Simon. Simon is in the Ubuntu repository
and acts as a user friendly front end for Julian.

http://sourceforge.net/projects/speech2text/

As a GUI simon can access a variety of dictionaries, Sphinx,
Julian, VoxForge and so on...

The forums for the simon project are well maintained and contain,
reasonable answers to neophyte and well as experienced users.
The GUI is years ahead of the samples I've been able to see
supplied by Sphinx or Julius so far.

For an intelligent discussion of the problems and current status of full dictation in existing open source solutions the simon/speech2text forum is informative.

http://sourceforge.net/projects/spee.../topic/3685883

simon is a slick and impressive GUI, while admittedly limited to the current speech to text resources it seems well poised to advance and also benefit from users feedback. It seems suitable for a limited commands dictionary but not full dictation, but is trainable, according to the author.