You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Looking at Dejavu code, it should not be very difficult to embed AdVent engine directly into a Dejavu "recognizer" plugin. It would still need to use threads, but would save CPU by reading the input stream once. Current AdVent engine treating entire Dejavu as a black box spends quite some time on creating and destroying pyAudio objects. Recognizer base class might need to be changed too, as it was not designed specifically for streaming use.
If implemented, it would require a major redesign of AdVent, so not earlier than the next major version.
The text was updated successfully, but these errors were encountered:
Looking at Dejavu code, it should not be very difficult to embed AdVent engine directly into a Dejavu "recognizer" plugin. It would still need to use threads, but would save CPU by reading the input stream once. Current AdVent engine treating entire Dejavu as a black box spends quite some time on creating and destroying pyAudio objects. Recognizer base class might need to be changed too, as it was not designed specifically for streaming use.
If implemented, it would require a major redesign of AdVent, so not earlier than the next major version.
The text was updated successfully, but these errors were encountered: