I've noticed that the wiki transcriptions for some of the recent Stack Overflow Podcasts are kind of weak. Clearly, this task calls for a computer program. Is transcribing audio to text (ideally with speaker labels so we know who said what) something that could feasibly be accomplished in software? Are there any active open-source software projects attempting to implement such functionality?
+3
A:
Believe me, I have searched for this before. There are slim to none text to speech that are open source or free to use. From my search there weren't any free speech to text synthesizers. These things are so hard to code and expensive that they can't really be made with an open source approach. If you really need this you would have to purchase it from a company. (although I don't know any off the top of my head).
Cory Walker
2009-08-14 02:21:35
+1
A:
I've looked into this a little. I tried the Microsoft Speech API but got very poor results. I've been wanting to look into the CMU Sphinx project, especially the Transcriber demo.
Jon Galloway
2009-09-18 17:16:52