Look here : https://www.kittenbot.cc/products/kittenbot-sugar-series-automatic-speech-recognition-module?_pos=4&_sid=cde2cce8f&_ss=r. It works without training, so it has a fixed vocabulary, found here : https://docs.google.com/spreadsheets/d/1jhytsqnPJN3KxcEcaBpT1_mlZgWySmRJ/edit#gid=503799978. I will use the four "colour lights" to light virtual leds on the Nextion and so trigger commands, and "speed up" to start the train. Unfortunately the "ambience" microphone is sensitive to "ambient" noise, this decreases the reliability. So I made 2 "hardware hacks", I removed the SMD led (with the Dremel ! ) because it could not display "blue", and the standard microphone and soldered two wires to a female jack, now it works fine with my Labtec headset. Three issues remain : first it communicates at 115200 bauds, this needs a special clock frequency so the Atmel can use this baudrate (I tested with an USB to serial adapter and the terminal program), second the feedback voice is absolutely horrible, like a character from a Japanese anime (listen to the attachment) and third the less stupid wake up expression is "Hello Kitty" (!). But for those people who think that the "cartoon interpretation" is too bad : Kitty is (like Amazon's "Alexa" ! ) a female "first name", it is a character in Leo Tolstoï's "Anna Karenina".
WARNING : normal commands are (red/green/yellow/blue) light on ; but if one says "red" alone, it can be interpreted as "Rest", which sets the module into idle mode, and there is no feedback for this ; and so You are surprised by getting no reaction ... until You say again "Hello Kitty" !
I thought first that it was the perfect system, recognition works fine and quickly, but the point I mentioned under "WARNING" is the "Achille's heel" ... especially because there is no feedback when system goes idle after the "rest" command ... So I wonder if I will REALLY integrate it into my layout. After all, 100 percent of the winners took a chance ... but not 100 percent of the people who take a chance will be winners !
:)
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Look here : https://www.kittenbot.cc/products/kittenbot-sugar-series-automatic-speech-recognition-module?_pos=4&_sid=cde2cce8f&_ss=r. It works without training, so it has a fixed vocabulary, found here : https://docs.google.com/spreadsheets/d/1jhytsqnPJN3KxcEcaBpT1_mlZgWySmRJ/edit#gid=503799978. I will use the four "colour lights" to light virtual leds on the Nextion and so trigger commands, and "speed up" to start the train. Unfortunately the "ambience" microphone is sensitive to "ambient" noise, this decreases the reliability. So I made 2 "hardware hacks", I removed the SMD led (with the Dremel ! ) because it could not display "blue", and the standard microphone and soldered two wires to a female jack, now it works fine with my Labtec headset. Three issues remain : first it communicates at 115200 bauds, this needs a special clock frequency so the Atmel can use this baudrate (I tested with an USB to serial adapter and the terminal program), second the feedback voice is absolutely horrible, like a character from a Japanese anime (listen to the attachment) and third the less stupid wake up expression is "Hello Kitty" (!). But for those people who think that the "cartoon interpretation" is too bad : Kitty is (like Amazon's "Alexa" ! ) a female "first name", it is a character in Leo Tolstoï's "Anna Karenina".
WARNING : normal commands are (red/green/yellow/blue) light on ; but if one says "red" alone, it can be interpreted as "Rest", which sets the module into idle mode, and there is no feedback for this ; and so You are surprised by getting no reaction ... until You say again "Hello Kitty" !
Last edit: Bertrand BAROTH 2022-11-12
Sounds like my wife Kitty. Sounds great!
https://www.kittenbot.cc/products/kittenbot-sugar-series-automatic-speech-recognition-module?_pos=4&_sid=cde2cce8f&_ss=r
I thought first that it was the perfect system, recognition works fine and quickly, but the point I mentioned under "WARNING" is the "Achille's heel" ... especially because there is no feedback when system goes idle after the "rest" command ... So I wonder if I will REALLY integrate it into my layout. After all, 100 percent of the winners took a chance ... but not 100 percent of the people who take a chance will be winners !
:)