#Echo in non-native integrations
sweet gate
HI @dull gale - ideally you can use our sdks that are built to de-echo the sound on the client. If you capture audio on your own, then unfortunatelly you are in charge of that for the moment.
I do wonder what is your usecase, is it a lacking support for your platform?
dull gale
I am using conversational api in a JVM app for our robot. Seems like 11labs js API is using audio source from webrtc inbuilt lib which has noise and echo cancellation, python api has pyaudio that seems to be cleaning up the stream.
However, in JVM there isn't anything easily available and if I use speaker with a mic, the agent reply goes back into mic and agent asumes it's the user speaking.
Anyways, I spent all day trying to solve I and the only solution that worked for me is to build a native webrtc c++ source code (which has inbuilt noise and echo cancellation) and then use JNI interface to get clean audio stream from that native library.
It's a bit annoying as I need yo rebuild the library if I deploy on a different platform, but at least it works. No echos ,no feedback loops, works like magic