- Messages sorted by: [ date ][ thread ][ subject ][ author ]
- Next message: Aaron Green: "Making MouseGrid more usable"
- Previous message: Susan Fulton: "Re: Voice-Users Digest V1 #492"
a recommended sound card, with lots of RAM and fast CPU speed should do
just fine. The new Dell desktops with the Montego card I have seen work
beautifully; ditto the Micron. Others as well.
The NC-80 is a pretty sorry microphone. VXI is a good microphone; if you
are using the battery box, that could be a source of problem. Also,
consistent positioning is critical, so run the audio setup wizard
frequently to make up for inconsistency. With VVGold, make corrections in
order, top to bottom. Correct single words; take advantage of Add Phrase.
In Nat correct phrases; take advantage of the Train button in the
correction window. Make a recording in the Sound Recorder and see what you
sound like.
At 01:50 PM 7/7/98 -0500, Russell Crawford wrote:
>Hopefully you folks can help me out. I have spent quite a bit of time with
>IBM's and Dragon's tech support, and they have been unable to find anything
>to explain the poor accuracy I'm getting. I have been using the both
>ViaVoice Gold and Dragon Naturally Speaking Preferred for about 3 months
>now on a daily basis. Despite trying all of the non-hardware related
>recommendations from the voice-users mailing list, like Vocabulary
>Expander, and always correcting everything and doing the full enrollment, I
>have been unable to get my recognition to anything better than 84% with
>ViaVoice. Dragon Naturally Speaking Preferred does only slightly better
>(86%) with about the same amount of work on the same things.
>
>Both IBM and Dragon's tech support say my microphone's signal to noise
>ratio and input levels are OK based on the software utilities that come
>with the products. Playback of my audio files with Sound Recorder doesn't
>reflect any electrical noise or anything else out of place I'm using a
>Crystal CS4232 sound card on a Toshiba Tecra laptop with 150 MHz processor
>and 144 Megabytes of RAM. I've tried both of the microphones that I have
>(one from Andrea, the NC-80, and one from Dragon which I assume is a VXI).
>Updating my sound card drivers did nothing but cause both programs to
>present errors when I try to start them. I borrowed a Thinkpad 700E for a
>week and trained it and obtained no benefit other than a lot faster refresh
>rate despite that IBM's web site says its certified as functioning well.
>
>My questions are: do these products really deliver 95%+ accuracy under
>optimal circumstances, is there anything else I could try to figure out why
>my Toshiba is so lousy, and what would be a machine configuration that
>would be essentially guaranteed to work if I were to buy a new desk top
>system since the conventional wisdom that laptops aren't very good at this
>has certainly been confirmed in my experience? It's been recommended to me
>that I get a preamp for my microphone by a friend, but I don't see why that
>would help if my input volume level is acceptable.
>
>Your assistance is greatly appreciated.
>
>Russell
>
>
>
>
------------------------------------------------------------
----SF/ using speech recognition software/forgive odd typos
- Next message: Aaron Green: "Making MouseGrid more usable"
- Previous message: Susan Fulton: "Re: Voice-Users Digest V1 #492"
 |  |