Look into Acoustic Echo Cancellation to prevent self-interruption

[SashaankG]

I added support for interrupt behavior, allowing a user to interrupt the robot while it is speaking. However, enabling this functionality introduced a new issue where the robot begins interrupting itself.

The problem occurs because the system’s speech detection pipeline cannot currently distinguish between user speech and the robot’s own TTS output. As a result, the audio produced by the robot is captured by the microphone and flagged as valid speech input.

Possible solution I am looking into/thinking about-

Acoustic echo cancellation

going to see how feasible this is for the jetson and implementations

https://forums.developer.nvidia.com/t/echo-feedback-issue-speaker-audio-being-captured-by-microphone-on-jetson-orin/354190/6

looking into solutions/ideas in this link

[notzabir]

[1] https://docs.nvidia.com/deeplearning/maxine/audio-effects-sdk/index.html

[2] https://docs.nvidia.com/deeplearning/riva/user-guide/docs/public/tutorials/asr-noise-augmentation.html