This message was deleted.

Publish audio is pointing at 16400 for audio also. Is that a typo?

Hi, I noticed you are using webmmux for VP8. the CLI doesn't currently support WebM container. It needs to be ivf container instead. give avmux_ivf a try instead: https://gstreamer.freedesktop.org/documentation/libav/avmux_ivf.html

^^ that was a typo: Hmm you mean like this?

gst-launch-1.0 rtspsrc location=<rtsp://someurl> latency=0 name=src \
    src. ! rtph264depay ! h264parse ! avdec_h264 ! vp8enc ! avmux_ivf ! tcpserversink port=16400 \
    src. ! decodebin ! audioconvert ! opusenc ! oggmux ! tcpserversink port=16402

I still get:

2022/10/09 17:54:21 "msg"="Could not start writing" "error"="IVF signature mismatch"

apologies I am a total newb at this

ivfreader is expecting a file header that is probably missing in this setup

Is that related to my gstreamer pipeline?

I suspect it's because ivfreader expects a file starting with the ivf header, but gstreamer isn't writing the header until the stream is finished

you might need to modify the cli to accept raw live vp8 streams. we have not tested it with live VP8 data. H.264 bitstreams should work fine tho. a number of folks are using that

Hmm I’ll see if I can figure that out and maybe I’ll make a PR. H.264 doesn’t work in an iOS simulator with WebRTC. I get a solid red box. I saw a handful of threads around that.

that's possible! we'd have to take a look. live stream function was contributed by the community. I'm not too familiar with it's limitations..

If you can point me to the commits or area to look at it will give me a good start 🙏🏼

This is where the file parsing takes place: https://github.com/livekit/server-sdk-go/blob/main/readersampleprovider.go#L139 Currently the only three container formats that we support are ogg, ivf, and ogg. Some digging would be needed to figure out if the hypothesis of "Gstreamer is writing proper header after the fact" is true

update, I have not made much progress on the VP8 issue but seeing if I can just work with H264 for now. The stream works well but the sound still doesn't come through. Curious if anyone has had any luck using tcpserversink for audio. I just installed the latest livekit-cli since I saw some recent commits regarding audio files. I can confirm I don't hear crackling noise like I used to but I am now just not getting any sound even though the mic becomes "active" I did notice this

mux

error in the logs.

2022-10-18T17:53:12.878-0400	ERROR	livekit.mux	mux/mux.go:112	mux: failed to read from packetio.Buffer short buffer	{"room": "my-first-room", "roomID": "RM_7jExbkYSsFYL", "participant": "bot", "pID": "PA_FSUhcKRk9qZR", "remote": false, "transport": "PUBLISHER"}
<http://github.com/pion/webrtc/v3/internal/mux.(*Mux).readLoop|github.com/pion/webrtc/v3/internal/mux.(*Mux).readLoop>
	<http://github.com/pion/webrtc/v3@v3.1.45/internal/mux/mux.go:112|github.com/pion/webrtc/v3@v3.1.45/internal/mux/mux.go:112>
2022-10-18T17:53:13.023-0400	INFO	livekit	rtc/participant.go:1231	sending offer	{"room": "my-first-room", "roomID": "RM_7jExbkYSsFYL", "participant": "web", "pID": "PA_AsW27vusn6qk", "remote": false, "transport": "SUBSCRIBER"}