Does VoiceAttack use the same speech engine as Cortana ?

[Agent86]

Does VoiceAttack use the same speech engine as Cortana ?

Just curious. VoiceAttack does not seem to like long voice commands and does better with a 3 or 4 word command.

Is there some reason for this ?

Please advise
Thanks

[Pfeil]

It does not, no.

Cortana uses an online engine, meaning it sends what you say to Microsoft's servers for processing. This allows for greater accuracy (as Microsoft, like Google and Amazon, has a lot of data collected from its users to compare to), but at a cost of latency.

VoiceAttack uses an offline engine, which requires no connection to the internet, but uses (in case of the built-in Speech Recognition Engine) a recognition model trained only using your voice
Speech Platform 11, which can optionally be installed and used by VoiceAttack, is also offline, but uses a static model (meaning it doesn't have to/can't be trained); The trade off it that it only allows for predefined commands, not dictation.


If you're having issues with recognition, have a look at the "VoiceAttack is listening, but what it hears is complete gibberish compared to what I said" section of this topic.

[Agent86]

It does not, no.

Cortana uses an online engine, meaning it sends what you say to Microsoft's servers for processing. This allows for greater accuracy (as Microsoft, like Google and Amazon, has a lot of data collected from its users to compare to), but at a cost of latency.

VoiceAttack uses an offline engine, which requires no connection to the internet, but uses (in case of the built-in Speech Recognition Engine) a recognition model trained only using your voice
Speech Platform 11, which can optionally be installed and used by VoiceAttack, is also offline, but uses a static model (meaning it doesn't have to/can't be trained); The trade off it that it only allows for predefined commands, not dictation.


If you're having issues with recognition, have a look at the "VoiceAttack is listening, but what it hears is complete gibberish compared to what I said" section of this topic.

Ok, thanks

I'm not really having too much recognition problems it seems to work quit well mostly.

It's just that the longer voice commands tend to get things wrong. Not always but accuracy with 3 words is almost 100% accurate while 4 words appears to decrease accuracy maybe 90% and the more words the less accurate and so on.

Can VoiceAttack use Cortana if desired ?

[Pfeil]

Can VoiceAttack use Cortana if desired ?

That is not currently possible, no.

[Agent86]

Can VoiceAttack use Cortana if desired ?

That is not currently possible, no.

Ok, thanks.
All working well, just curious about it.