1 person found this review helpful
Not Recommended
0.0 hrs last two weeks / 0.2 hrs on record
Posted: Mar 15 @ 1:54pm
Updated: Mar 15 @ 10:37pm

I did the training a couple times and even after that it seemed to be very hit or miss with recognizing very simple words.
"say" kept turning into "They".
"open" was "the open" or even "all they can".
I had a feeling I was wasting my time with this so just requested a refund.

EDIT: Thank you to the dev who replied quickly. I might play around with the demo and see if I can do some adjustments and get things to work better. But at this point I'm not to thrilled with the reliance on Windows very imperfect voice recognition. I guess also I feel like this being 2024, we're getting well into ai voice recognition, I feel like this should work much better out of the box.
Was this review helpful? Yes No Funny Award
Developer response:
gmagenheimer  [developer] Posted: Mar 16 @ 5:38pm
Edit (reply to updated review) - Normally, Windows' speech engine works surprisingly well out of the box using user-defined commands. Question - did you add any commands (spoken phrases)? If you just turn VA on and start speaking without any pre-defined phrases, the speech engine is basically dumb as it has nothing to go by. If you did add phrases and nothing is recognized, there's a really good chance that there's something up in your audio chain. Come on by the VA Discord and have a chat (I stand by the reliability of this app - not because I'm the dev, but because I use it constantly).

On a side note, I've tried to integrate more current, service-based speech recognition into VoiceAttack, only to find the performance in this app context to be abysmally (I mean seriously) slow. As it stands, the speech recognition built into Windows (for this app/context) is actually quite good and relatively fast to respond (and least of all, free). Once we get to a point where the performance of the service-based speech recognition can be on-par with what Windows offers, I'll look at making the transition. In the meantime, I'm thinking about making one or more service-based speech recognition systems available as alternatives, just to give users a variety (even if they are paid services).

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

It's unfortunate that you had to walk away on VoiceAttack - normally, issues related to the speech engine can be rectified with a little bit of tweaking to get it just right (even after training). With VoiceAttack, it's a two-way street - VoiceAttack can do its part and work wonderfully if the local PC environment is configured a certain way. Usually out of the box, VoiceAttack will work straight away, but if there's something out of alignment in the local environment, some adjustments will be needed. Things like the microphone being too 'hot' or line noise - there has even been (multiple) cases where a webcam's mic was inadvertently being used as the speech engine's recording device. A quick stop by the VoiceAttack Discord or user forums (or Steam discussions) is usually all it takes. Anyway, if you change your mind and want to try again, there's always the fully-functioning demo version of VoiceAttack available on Steam that can be used to work with to get everything up to working order.
Comments are disabled for this review.