loowee
Member
Hello there cocoonners !!
NOTE : This post is in regard to ALL foreign languages speaking people using homeseer. ( Foreign = not english )
First thing :
Basically, I have been trying to find some way ( scripting / modifying HS config files / etc ) to translate the default trigger commands such as "TURN ON" or "GET ME" or "WHAT IS THE STATUS" to french in homesser setings.
Vista's french VR engine ( 8.0 ) works perfectly and is as good as english for french recognition. French TTS from loquendo is also a perfect voice and works perfectly with my setups.
I conclude the only way to get HS to understand french trigger phrases is to disable these default trigggers with some scripting and force all devices VR commands to have the french terms for "turn on / ouvrir" for example. This limits VR possibilities extremely and I fear adding so many terms in voice commands for devices or events will make the whole VR understand very badly.
Secondly :
Is thhere any way to increase VR reliability when using it with multiple devices and events, basically, when using a huge word dictionnary ? I use the acoustic magic voice tracker and find the recording is just excellent when using with with windows VR directly. I dont understand why windows can decode almost up to 99% of a 300 word text and mmy homeseer nistall, having just an approx. 50-60 words disctionnary will practically invent things I did not say. The reliability is less than 20% right now.
DO WE REALLY HAVE TO USE SPECIFIC DICTIONNARY SCRIPTING TO GET ANY KIND OF RELIABILITY ? This method works well as it limits the VR word counts but takes a huge time to load in between the switches.
Finally :
Isnt there any software or filter to cancel the noise generated by the same machine that records the VR ? Sound cards are now very powerfull these days and since it can modify voice live, I do not see why it could not just cancel music it is outputting on the mic input. Basically filtering the frequencies that the music / noise generates, intelligently that is so it does not cut voice on same freq.
I think this would be the inverse of a karaoke as it would filter music or other freq. from input and keep voice most common freq. active. That been said, I not even sure how karaoke does the job so well, hearing just a FAINT voice while music is not affected at all.
Well, I just hope someone is as interested as me in using the endless VR possibilities that a reliable combo ( mic - VR engine ) would offer. Might just be a matter of time but Im pretty confident the technology is ready just by using Vista's VR ( french or english ) a few times for test purpose.
Thx guys !
NOTE : This post is in regard to ALL foreign languages speaking people using homeseer. ( Foreign = not english )
First thing :
Basically, I have been trying to find some way ( scripting / modifying HS config files / etc ) to translate the default trigger commands such as "TURN ON" or "GET ME" or "WHAT IS THE STATUS" to french in homesser setings.
Vista's french VR engine ( 8.0 ) works perfectly and is as good as english for french recognition. French TTS from loquendo is also a perfect voice and works perfectly with my setups.
I conclude the only way to get HS to understand french trigger phrases is to disable these default trigggers with some scripting and force all devices VR commands to have the french terms for "turn on / ouvrir" for example. This limits VR possibilities extremely and I fear adding so many terms in voice commands for devices or events will make the whole VR understand very badly.
Secondly :
Is thhere any way to increase VR reliability when using it with multiple devices and events, basically, when using a huge word dictionnary ? I use the acoustic magic voice tracker and find the recording is just excellent when using with with windows VR directly. I dont understand why windows can decode almost up to 99% of a 300 word text and mmy homeseer nistall, having just an approx. 50-60 words disctionnary will practically invent things I did not say. The reliability is less than 20% right now.
DO WE REALLY HAVE TO USE SPECIFIC DICTIONNARY SCRIPTING TO GET ANY KIND OF RELIABILITY ? This method works well as it limits the VR word counts but takes a huge time to load in between the switches.
Finally :
Isnt there any software or filter to cancel the noise generated by the same machine that records the VR ? Sound cards are now very powerfull these days and since it can modify voice live, I do not see why it could not just cancel music it is outputting on the mic input. Basically filtering the frequencies that the music / noise generates, intelligently that is so it does not cut voice on same freq.
I think this would be the inverse of a karaoke as it would filter music or other freq. from input and keep voice most common freq. active. That been said, I not even sure how karaoke does the job so well, hearing just a FAINT voice while music is not affected at all.
Well, I just hope someone is as interested as me in using the endless VR possibilities that a reliable combo ( mic - VR engine ) would offer. Might just be a matter of time but Im pretty confident the technology is ready just by using Vista's VR ( french or english ) a few times for test purpose.
Thx guys !
