Voice Recognition Software

[skymama 35]

This is funny, my boss just required me to get on the forums and ask you skydivers a question. I wonder what would happen if I said no?

Anyway, he wants to know if anyone has had success using voice recognition software. Does it work any better/different on a laptop vs. a desktop?

After you spend hours getting one computer to recognize your voice, is there a shortcut to make it work on another one, or do you have to go through the whole thing again?

She is Da Man, and you better not mess with Da Man,
because she will lay some keepdown on you faster than, well, really fast. ~Billvon

[TFFTM 1]

***Limited experience with lower end software.***

Its the Mic that will make the diffrence not the hardware platform.
If the software allows for multiple profiles, you can copy a profile from one machine to another rather than "Retrain" each machine. The more you train your profile the better it will perform.

The software I used to use (Dragon speak I think) toook several samples of common words and phrases and would overlay them. It would then search for a matching pattern.
Unfortunatley I mumble alot and had to retrain several times. I was the cause of many hours of laughter, me sitting yelling at my computer that "I said X and not Y."
One of the upsides is it did teach me to innunciate better in casual conversation.

Good luck

edit to add: Its sometimes more about training yourself to speak clearly than training the software to learn your speech patterns.

BSBD
Home of the Alabama Gang

[ScottishJohn 25]

I tried via voice for a while but it took ages for it to learn my voice (aparently I have an accent) but even when I had trained it up to recognise my voice I could still type faster than the software could.

Things have changed since then so it might be more reliable.

Would it work in an open plan office with everyone talking ?

----------------------------------------------------------------------
If you think my attitude stinks you should smell my fingers

[TFFTM 1]

One more thing to add. When transferring profiles remember that whatever background noise is in the original profile will be copied as well. So using the profile in a different noise environment will affect its ability to recognize.

BSBD
Home of the Alabama Gang

[BillyVance 34]

I played with a simple program like that years ago at work where we had those Compaq 400 mhz computers and the keyboards had a small slit in it for a microphone. We could be in Windows and say "file" or "start", and follow the menu options and so forth and be able to open/close applications, etc...

I'm not sure but on a lark and for shits and giggles, I said "fuck you" to the computer. It froze up on me...

That shit fucker...

"Mediocre people don't like high achievers, and high achievers don't like mediocre people." - SIX TIME National Champion coach Nick Saban

[Remster 24]

Are you asking about voice recognition, or speech recognition?

Voice is when a system will recognizwe your specific voice (think security).

Speech rec is when you can give a dictation to your PC

Remster

[SudsyFist 0]

Smartie pants.

[BillyVance 34]

Quote

Are you asking about voice recognition, or speech recognition?

Voice is when a system will recognizwe your specific voice (think security).

Speech rec is when you can give a dictation to your PC

You're right... it was speech rec software I was playing with, not voice rec...

edited to add: I don't think they had voice rec technology widely available 9 years ago.

"Mediocre people don't like high achievers, and high achievers don't like mediocre people." - SIX TIME National Champion coach Nick Saban

[Remster 24]

Quote

Smartie pants.

No.. I used to work with both technologies... I'm not just a pretty face

Remster

[ematteo 0]

Dragon Naturally Speaking is the leader in pc-based dictation (and pc command and control) software.

http://www.nuance.com/

My cousin is a quadripalegic and has used this for about 15 years successfully.

Dragon's product is based on statistical language modeling, and it learns your specific voice over time. The model is transferrable to a new pc, and can supposedly work with multiple pcs on a network.

There are business-oriented systems out there that work without training (call UPS and track a package, or American Airlines and check flight status) but they have limited vocabulary and use a different type of model.

Nuance (aka Dragon, Lernaut and Hauspie, Speechworks), IBM and Phillips are the major providers of English-language speech rec.

Evan

[ematteo 0]

For what it's worth, you can combine both features. Nuance made (probably still makes) a voice verification app that ran in combination with their speech rec. Useful for financial services apps. Especially good for reducing false negatives, since you can combine biometric and knowledge authentication.

[hobie331 0]

Just wondering if there have been any huge updates with such software. Thinking about trying it and getting rid of the cost of paying a transcriptionist. It's been a couple years or so since I talked with anyone using it and at the time they were not satisfied and had moved on, or maybe back, to transcription.

I'm really proud of myself today. I actually went looking for a thread on this subject instead of starting a new one.

It's a gift, I don't try to explain it.

[skymama 35]

Quote

I'm really proud of myself today. I actually went looking for a thread on this subject instead of starting a new one.

I'll give you a sticker for that.

She is Da Man, and you better not mess with Da Man,
because she will lay some keepdown on you faster than, well, really fast. ~Billvon

[mdrejhon 8]

Quote

Just wondering if there have been any huge updates with such software. Thinking about trying it and getting rid of the cost of paying a transcriptionist. It's been a couple years or so since I talked with anyone using it and at the time they were not satisfied and had moved on, or maybe back, to transcription.

Although I'm deaf, I keep an eye on this as transcription software may help others put their words into text for me.

The latest version of the transcription software have improved quite a lot, and requires minimal training (sometimes no training at all) and it learns on the fly. So if you are still using transcription software, then upgrade to the latest. Just make sure you have a good noise-cancelling headset. It could even be a high end Bluetooth headset, for convenience.

[SuFantasma 0]

Well, I take it your boss doesn't have any connections at the CIA ... they have quite an impressive setup for speaker-independent voice recognition as well as message heuristics.

What is it that your boss wants to do with it ? Just dictate memos? Control the operating system? The intended use is more important than any other consideration.

Y yo, pa' vivir con miedo, prefiero morir sonriendo, con el recuerdo vivo".
- Ruben Blades, "Adan Garcia"