I have a project to achieve a simple speech recognition on a Raspberry Pi with Pocketsphinx. The workflow I imagine is the following :
I say a keyword to make the Raspberry Pi listening "Pi" for instance. A led connected on the Raspberry Pi turns into a given color to show that the Raspberry Pi is listening for further instructions (during 1 or 2 secondes for instance).
The initial "PI" keyword is mandatory to avoid false positives for further instruction.
I will then give my instructions :
"Turn on the lamp" --> ssh command to turn on the lamp
"Pause video" --> ssh command to pause the video etc...
Instructions would be predifined by myself.
The whole thing will be in French language and not english. I just want to know if something like that is possible before spending too much time in the documention and find out that it's actually not possible.
Thanks
Last edit: ParanoYak 2015-11-17
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
It is possible, should work ok with latest french ptm models, you can check Android demo on how it works. You first listen in keyword spotting mode, then switch to grammar mode for command recognition.
The activation phrase "Pi" is too short though, we recommend 3-4 syllables, like "Votre Excellence".
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello,
I have a project to achieve a simple speech recognition on a Raspberry Pi with Pocketsphinx. The workflow I imagine is the following :
I say a keyword to make the Raspberry Pi listening "Pi" for instance. A led connected on the Raspberry Pi turns into a given color to show that the Raspberry Pi is listening for further instructions (during 1 or 2 secondes for instance).
The initial "PI" keyword is mandatory to avoid false positives for further instruction.
I will then give my instructions :
"Turn on the lamp" --> ssh command to turn on the lamp
"Pause video" --> ssh command to pause the video etc...
Instructions would be predifined by myself.
The whole thing will be in French language and not english. I just want to know if something like that is possible before spending too much time in the documention and find out that it's actually not possible.
Thanks
Last edit: ParanoYak 2015-11-17
It is possible, should work ok with latest french ptm models, you can check Android demo on how it works. You first listen in keyword spotting mode, then switch to grammar mode for command recognition.
The activation phrase "Pi" is too short though, we recommend 3-4 syllables, like "Votre Excellence".
ok thank you for the advice, "Votre excellence" sounds good ;)
I will dig into it to see if I can get the result I want.