Issues with Echo Cancellation

[yash-samespace]

I’ve integrated this library into a Tauri-based Rust project that leverages WebRTC technology and I’m encountering issues with the echo cancellation feature.

Here are the details:

• Environment: I built the library on macOS using meson build, following the instructions in the repository’s README.
• Setup: I initialized the audio processor and connected process_capture_frame and process_render_frame to the audio input and output streams provided by CPAL (audio I/O library).
• Testing: After calling process_capture_frame(), I used get_stats() to monitor the performance. I observe that has_echo is set to true in certain cases, which suggests that echo detection is partially working. However, there is still an audible echo on the remote side, and in other cases, it seems that echo isn’t detected at all.

To clarify my understanding: is the expectation that the raw audio stream passed to process_capture_frame() is directly modified to filter out the echo? Or is there another step I may be overlooking in the echo cancellation process?

Any guidance you could provide would be greatly appreciated. Please let me know if you need additional information on my implementation.

Thank you for your work and support

[bschwind]

Hi @yash-samespace ,

To clarify my understanding: is the expectation that the raw audio stream passed to process_capture_frame() is directly modified to filter out the echo?

Yes, it should be modifying the captured audio data in-place. Same with process_render_frame.

Have you been able to run any of the examples without issue? The karaoke example might be the most helpful in showing the expected usage of this library.