Open G10DRAS opened 5 years ago
How did you made hotword robocop?
what is the process of it?
For hotwortd detection I customized this project https://github.com/G10DRAS/dejavu-Hotword-Detection
so how you made a hotword? you are just recording the hotword yourself and you are storing the hotword in dejavu and then dejavu matches the hotword and activate the robocop?
so when i speak to robocop, what is the process from activation to speach to text?
so when i speak, first dejavu match the hotwork and activate the robocop and then you do speach to text with pocketsphinx and then execute the action?
so how you made a hotword? you are just recording the hotword yourself and you are storing the hotword in dejavu and then dejavu matches the hotword and activate the robocop?
-- Yeah record myself as well as others voice for hotword, then there will be post processing of voice like pitch bend, tempo change etc....
so when i speak to robocop, what is the process from activation to speach to text? so when i speak, first dejavu match the hotwork and activate the robocop and then you do speach to text with pocketsphinx and then execute the action?
-- Exactly, so no speech will go out of device.
thank you very much!
but pocketsphinx already has keyword spotting mode, i think we we can use for activation? they why to use dejavu?
RoboCop will recognize faces and greet, no need to utter Hotword/Keyword. RoboCop will recognize only pre-trained faces.