32 replies to this topic

[eternaltraveler]

speech to text software really has come a long way. Currently I am using speech to text software to write this very message. I remember trying out speech to text software a few years ago, and even after extensive training of the software, it still wasn't very effective.

But now, the results speak for themselves. It understands even technical subjects. For example reading from the neuroscience notes I have that in front of me, "a single presynaptic impulse releases glutamate into the cleft where it binds AMPA and NMDA receptors in the postsynaptic membrane. The AMPA receptor cation channel opens in the ensuing entry of sodium ions produce an end plate synaptic potential". blah blah

you get the point.

I guess I just like my new toy )

[Mind]

Cool....now the important things... how much? How long to train it? Is it faster than typing?

[Brainbox]

Impressive indeed. Can you tell some more about it, like affordability?

Edit: Do you have any experience yet with background noise?

Btw. did the software also handle the smiley... [!]

Edited by brainbox, 05 May 2007 - 02:42 PM.

[Live Forever]

What about thought-to-text? If they can get that hammered out, I'm in.

[Brainbox]

In my case that would be dangerous, very dangerous... [lol]

What about thought-to-text? If they can get that hammered out, I'm in.

[siberia]

What's the name of the product you're testing? I'm thinking of maybe using it as a way to better my pronounciation.

[Ghostrider]

It also depends on the quality of your microphone and sound card. I tired IBM ViaVoice back in 1998 and it was utter crap.

[zoolander]

Come on elrond you bastard! What's the name of the software

[zoolander]

It's got to be Dragon Naturallyspeaking 8 or 9. They have a medical suite

[eternaltraveler]

It's got to be Dragon Naturallyspeaking 8 or 9. They have a medical suite

correct

dragon naturally speaking 9

[eternaltraveler]

Impressive indeed. Can you tell some more about it, like affordability?

Edit: Do you have any experience yet with background noise?

Btw. did the software also handle the smiley...

in a way it handled the smiley. I just told it semicolon/closed parenthesis

affordability... heh, it's not mine, it's a friend's of mine, i believe it's about 200 bucks.

[eternaltraveler]

How long to train it? Is it faster than typing?

training? about 20 minutes.

I don't know if it's faster than typing per se, but the primarily limitation on my typing is my thought process, not the typing itself.

[eternaltraveler]

What's the name of the product you're testing? I'm thinking of maybe using it as a way to better my pronounciation.

you do have to be clearer than normal (but not ridiculously so). so it might help in that regard.

[Ghostrider]

training? about 20 minutes.

I don't know if it's faster than typing per se, but the primarily limitation on my typing is my thought process, not the typing itself.

hehe, I can sometimes type much faster than I think...in fact, I often do.

[modelcadet]

Yeah, I'm seriously debating buying some voice recognition software... my friend developed carpal tunnel syndrome, and I'm very soon going to be doing similar work. I'm thinking this, plus Jeff Han's multi-touch screen (which you can make yourself!), plus some sort of brain-hacking via electrical stimulation of specific data (computer statistics, audio information, etc)... I will be the ultimate machine!

[futureofscience]

I think it might be useful for me too. I am a terrible writer but very good when it comes to speaking so it would be beneficial.

And would stop my sup getting on at me to update my report [lol]

[futureofscience]

Just as a question, which version are you using? There is standard, preferred and professional. Is the standard version okay ?

[eternaltraveler]

preferred I believe.

Professional is like a thousand bucks

[Live Forever]

A quick search turns up lots of torrents of the software, if one were so inclined... [wis]

[futureofscience]

I've ordered a cheap copy off ebay with a headset included as I don't actually have one. Here's hoping it improves productivity and reduces nagging!

[Shepard]

Now I can post and get headshots at the same time.

Boom, headshot.

[futureofscience]

Well, I got my copy this morning and I'm really impressed, have already dictated a load of notes up. Still needs a wee bit of work but I'm very impressed I have to say [thumb]

[sentinel]

How much was yours FoS? I have never got round to teaching myself to touch type and this could be a good alternative..

Also, someone mentioned the background noise factor - does it cause much interference if people are speaking in the background as I work in an open-plan office...

Sentinel

[futureofscience]

It tends to have worked well even with some background noise, it's just a case of training it sufficiently, but I'm not sure as to using it in an office.

I got mine for 50 quid from Ebay (for the preferred version) with a headset thrown in as I didn't have one. It's also recognised quite a few "technical" words I wouldn't have expected.

[sentinel]

thanks - I might give it a go!

[Live Forever]

There are lots of torrents of it too:
http://www.torrentsp...ng_v9_Preferred
http://www.torrentsp...lly_Speaking_v9

[JonesGuy]

I'm currently trying to get a group of lawyers to use software like this, instead of dictating letters for their secretaries to type out. There is actually quite a bit of resistance to the idea, even though it would free up hundreds of hours of time per year at minimal opportunity cost

[modelcadet]

The implications of speech to text software are huge. Not just for dictation, but also for providing metadata for audio and video.

I'm currently -slowly- working on an open-source project called ReVoxer which will provide tools for dictation (especially for public domain audio books). There are so many people recording voice with text already present, it's much easier to provide time-stamping for individual words. The goal is to provide early use for the next wave in audio/video through public domain books.

My favourite implication is text karaoke, where a program highlights individual words as they are read in the audio file (for young readers, language learners, and the learning disabled).

Unfortunately, Naturally Speaking is the only real good software. The open source engine, Sphinx, still has a ways to go. Maybe more contributions to VoxForge or a similar site will help in this respect.

[Brainbox]

I'm currently trying to get a group of lawyers to use software like this, instead of dictating letters for their secretaries to type out.  There is actually quite a bit of resistance to the idea, even though it would free up hundreds of hours of time per year at minimal opportunity cost

I'm not resisted to secretaries either.

[basho]

I'm currently trying to get a group of lawyers to use software like this, instead of dictating letters for their secretaries to type out.  There is actually quite a bit of resistance to the idea, even though it would free up hundreds of hours of time per year at minimal opportunity cost

Secretaries probably do more than simply type out dictated letters. They may correct mistakes, follow up on inconsistencies, edit out cursing, format things nicely, sit around looking pretty, etc. All very important things.