Reolink Lumus Pro 4K – RTSP backchannel audio playback is garbled/choppy (PCMU/8000)

[Glimstara]

I’m trying to use go2rtc’s “play audio” / RTSP backchannel feature with a Reolink Lumus Pro 4K camera that supports two-way audio via ONVIF backchannel.

While the feature technically works, audio playback is extremely garbled, choppy, and sometimes at the wrong speed, to the point of being barely usable.

I’ve attached a phone recording (audio.wav) of what the camera speaker outputs when go2rtc plays a simple audio file. As you can hear, speech is unintelligible, unevenly paced, and heavily distorted.

I've tried .mp3 and .wav files, files preconverted to mono / 8000 Hz / pcm_mulaw, playback via go2rtc Web UI (“play audio”) and Home Assistant WebRTC Camera virtual media_player, and direct RTSP stream vs ONVIF-derived stream. All attempts result in similarly garbled, choppy playback.

go2rtc info & .yaml:

platform=linux/amd64 revision=dfe4755 version=1.9.13

streams:
  reolink_front_door: rtsp://admin:***@192.168.10.54:554/Preview_01_main

audio.wav

Any advice would be greatly appreciated.

[AlexxIT]

I don't even know how to debug this without having a camera. Even remote access won't help here.
The camera may require a special packet structure. And delays between sent data. This requires experimentation in the code.