This may be edited someday to have a point.
There has been some talk on one of the threads in a group for CCACaptioning (for captioning advocacy) about voice
recognition and CART, and whether voice recognition technology can become more applicable for daily situations, such as restaurants or meetings.
I did voice recognition CART for Calvary Church for a year and a half before we got a budget for a *real* CART pro (aka Karyn, Tennessee Captioning).
Voice Recognition captioning is generally done by one speaker who has trained the voice recognition software to his voice and style. Using a headset, he will repeat verbatim what is being spoken, into the mic.
I was using CaptionMic software http://www.mhsa.us/mhsa_cm.html (see videos of Captionmic in action: http://www.youtube.com/user/captionmic), sold by Mark Hall, who I believe is also a member of that list. CaptionMic is based on an embedded version of IBM’s ViaVoice technology. It’s a special application, because it delivers the text in a different way, decidedly more pleasant than normal dictation software, like Dragon, and it enables creating line 21 captioning.
I have also used Dragon Naturally Speaking, version 10, for the PC. I use it for creating transcripts of longer videos. Even without training, I can get a pretty good transcript.
There are a couple of points that are important to note about voice recognition.
1. Voice recognition is more accurate with a high quality headset and noise cancelling microphone.
2. Voice recognition accuracy is better with a trained speaker, someone who is conscious about not slurring.
3. Voice recognition accuracy is better with a trained speaker who has trained the software to his/her voice.
From a recent interview I heard on This American Life, I’m guessing that voice recognition technology faces the some of same
drawbacks that CI technology does. It can’t perform well in any situation that anyone with electronic hearing devices would have trouble. It really can’t do well with multiple voices, and has no way to filter out ambient noise, outside of what the microphone noise filtering can do.
With those caveats….
-Windows has built in Voice Recognition ->Setup Speech Recognition in Windows 7.
-Web based Chat Room with moderator capabilities. (I’m thinking of creating one setup for this)
Pro/Cons: +Cheap, +Better than lipreading, -Timing of the captions, strange delays
-Use Dragon for Speech Recognition
-Use Streamtext.net for broadcast
Pro/Cons: +better recognition/accuracy, =still inexpensive, -still has strange delays
-Use pro software like CaptionMic. I know there are others.
-Use a CART professional. Read Voice CART versus Steno CART here:
One the things I like to say about the voice CART: It was a “gateway drug” to having professional CART.
It’s better than not having something, but it’s nowhere near the quality we are getting now with an excellent steno CART.