I think there should be a client-Function ts3client_getRecordingBuffer which returns the FMOD-Sound-Buffer that is used to capture the Mic-Input.
This way you can Stop TS to record the mic-data (with ts3client_stopVoiceRecording() ) and insert something else into the buffer. It would make Music-Bots easier for example.
Or you can start recording yourself and add an echo to the mic-input.
Perhaps you would also need a new event onRecordingBufferChange that is called when Teamspeak changes it Recording Buffer.
I think it would be nice to have that feature.
Greetings

FH