Open Benjamin-Loison opened 6 months ago
To reduce workload may work like other well-known vocal assistants using a common prefix like Dis Siri or Ok Google. It may reduce workload by only listening carefully once have pronounced this prefix which may be detected in a more approximative fashion (note that can run precise listening to treat again the recorded prefix to make sure that it is not a false positive). The prefix could be Dis Termux. Should play a sound once have quickly or/and carefully detected the prefix to start speaking more precisely or depending if already listening with high quality but not parsing speech, then can not have such a sound, hence a delay to start giving an order.
Related to:
-----BEGIN PGP MESSAGE-----
hF4DTQa9Wom5MBgSAQdAPNH8MtM2jl4DOkkZc+AAXH7qLVNb09NIi1rHzz2LnXsw
YOauJPpJtUjmHLZYEk4OS1TygkPDLfr2goKM79+RLJk9MgAvP5yGfDgRiPjG3c7K
0nsBqkaba+vedDP9SIQV5u/H+MdlQFKil387YOAXGcBVwJnCCjD8CBBcia7cryaa
CknUNbGx/OvwT2znfFg8ha+Jcdn2oS8u+orVwnvF1Y8/kcpSNtOwRwXZ9Pa7l1zh
6GoUOVAQ+AqjSOgvWTBFZefW1JEXD77/dpgFOhU=
=32qL
-----END PGP MESSAGE-----
Related to:
-----BEGIN PGP MESSAGE-----
hF4DTQa9Wom5MBgSAQdA4lra9Ow79biS9ICHiKLa8gvmHX/cL6Zp9mg2k+5/9Wow
KFYYSlvoCZ3S+3HpQpyUB4lUhJwxp9ktfkTFDYzHfxbuGTsIyXIGCfW+PVqlOmDf
0nsBmwPvqLLnM7dgUn63PgZ2jol+S6PKzthd2tTNMp0Qtx4PT3kIQdsfwSjGR7Hu
jKMA8Y7cYlP7wmLRGbIGSGgCT14ts4GUr3eZXWp7jXq6lx4pxqh1YKCZg4y+t+yl
I9Hm96pvURQrcQb4wCPnUKhKtkZJN2Kvxxppoi0=
=0T6H
-----END PGP MESSAGE-----
Maybe if there is an audio being played can subtract it somehow to the audio recorded with the microphone.
Related to Benjamin_Loison/mobicoop-platform/issues/26.
Related to Benjamin-Loison/cinnamon/issues/66.
whisper_mic (690 stars) Source: whisper/discussions/75
looks interesting and there is no much code so can review it.
Only pyaudio
looks suspicious in requirements.txt
:
whisper_mic/blob/8e700b75a152ee36db5fdd16af125cea8f9843a8/requirements.txt#L7 Except above requirement, whisper_mic/blob/8e700b75a152ee36db5fdd16af125cea8f9843a8/pyproject.toml looks legitimate.
Have not checked other files yet.
able (38 stars) looks less interesting.
Related to Benjamin-Loison/termux-app/issues/{37,38}.
Related to Benjamin_Loison/Food/issues/44.
For temporary use-cases:
Can use Termux and Whisper to do so but may have load issue, could make one of my server execute the speech recognition but then pay attention to security/privacy issues.
Investigate www.home-assistant.io.
The relative being:
Related to Benjamin-Loison/firefox-android/issues/16 and Improve_websites_thanks_to_open_source/issues/154.
+80