TenTec
[Top] [All Lists]

Re: [TenTec] Speech to Text

To: Discussion of Ten-Tec Equipment <tentec@contesting.com>
Subject: Re: [TenTec] Speech to Text
From: mike bryce <prosolar@sssnet.com>
Reply-to: Discussion of Ten-Tec Equipment <tentec@contesting.com>
Date: Mon, 6 Dec 2010 11:33:29 -0500
List-post: <tentec@contesting.com">mailto:tentec@contesting.com>
I tried to use Macspeech Dictate (powered by dragon) and have had lack luster 
results.

perhaps my speech is less than perfect, but I can't get it to do my call sign 
correctly.

At $300 bucks, microphone included, I was hoping for better results.

the claim 99% accuracy. I find, my voice, I'm lucky to break 70%

even at 99% that's one word in ten and 10 words per hundred. That's a lot of 
corrections to be made.

and since you have to train the software to your voice, I don't believe it 
would work on signals coming in from the radio.


just my .02 worth.

mike

Mike, WB8VGE
SunLight Energy Systems
The Heathkit Shop
http://www.theheathkitshop.com/
J e e p
o|||||||o 
If it wasn't for great distances, we would all live a lot closer!




On Dec 6, 2010, at 10:45 AM, Richards wrote:

> I have every commercially available speech recognition program published 
> since 1992, and have been a beta tester for every manufacturer (except 
> Phillips who pulled out of the North Am market ten years ago...).  I 
> have been a product tester, reviewer and Guinea Pig for microphones and 
> related input devices for several vendors.
> 
> My recommendation is to get a headset mic - which moves around with you 
> as you turn or move your head, and this helps keep the sound volume and 
> all consistent.
> 
> I also recommend you try  1)  Dragon NaturallySpeaking as the most 
> accurate speech recognition engine at cost, but if cost is a barrier, 
> then  2) Microsoft Speech which comes free within various recent 
> iterations of the Windows OS and some recent versions of MS Word.
> 
> There are valid differences of opinion - different users get different 
> quality results with various speech engines - but I much prefer Dragon 
> NatSpeak as I think it is the more accurate speech engine, and it has a 
> refined UI, whereas MS Speech is still in first beta (has been always - 
> probably will never become finished...) and lacks the full UI that the 
> Nuance (Dragon) product has.   It just is a more complete experience and 
> there is more formalized support for it than for the one built into the 
> OS- which is an "undocumented" feature.
> 
> You can purchase some third party add-ons to the Dragon product, 
> including one called KnowBrainer(R) which is a set of pre-defined 
> computer program control commands.   This makes the programming of 
> commands (as opposed to just straight dictation of words) easier and is 
> more robust than the native available command structure that is built 
> into the program.   Such third party add ons are not as plentiful for 
> the MS Speech recognizer.
> 
> Get a better headset (but don't spend a lot of money) on a good 
> microphone as the one in the box is just OK -- it will do fine, but a 
> slightly higher expense is warranted.   Do not spend a lot of money on a 
> microphone - many vendors will make you think you need that -just like 
> there are guys who say you need a $400 gold sputtered large diaphragm 
> studio condenser mic, with noise gate, compressor, equalizer, and all 
> the other trimmings for ham radio.   In fact, I can explain why having a 
> studio quality microphone for speech recognition can actually degrade 
> performance  (hint:  think sound card sample rates....)
> 

_______________________________________________
TenTec mailing list
TenTec@contesting.com
http://lists.contesting.com/mailman/listinfo/tentec

<Prev in Thread] Current Thread [Next in Thread>