I followed the README instructions in Zoom Meeting SDK Linux Sample App.
I build the image using the command mentioned in README:-
docker build -t msdk-5.17.5-on-ubuntu-compact -f Dockerfile-Ubuntu/Dockerfile .
and run the container using the command :- docker run -it --rm msdk-5.17.5-on-ubuntu-compact
After running the container, at last I got “Failed to connect the meeting server” message in terminal . Please find the attached screenshot of the terminal.
@chunsiong.zoom
After removing recording token , LinuxChun was able to join the meeting and I got the
following output in terminal .
Please find the attached screenshot of terminal.
After changing the recording token , Bot was able to join the meeting but I’m not able to get raw video and audio data. Terminal processing is not continuing. Please find the attached screenshot of terminal.
@chunsiong.zoom
Recording token is expiring within 120 sec. So I tried to build image and run container within 120sec.
Bot is able to join the meeting and also recording is happening successfully. Please find the attached screenshot below.
Please guide me where the raw video and audio data will be saved ?
@chunsiong.zoom@donte.zoom
Recording token that I’m getting by calling this API : GET /meetings/4174859488/jointoken/local_recording
is expiring in 120sec. By the time container spins up, recording token is expiring, so Bot is not able to join the meeting.
Only once bot was able to join the meeting when container got spinned up within 120sec. Later on container is taking more time to spin up and so bot unable to join the meeting.
Please suggest what can we do about this ?
@abhishekmv.dev , this is beyond the scope of this SDK as it is best practices on docker containers.
The creation of docker container is typically done once.
Subsequently when spinning it up, you do not need to build again as the image has already been built.
You just need to pass in the new auth token, meeting number, password and recording token.
Thank you @chunsiong.zoom . I’m able to get output.yuv file.
I’m using ffmpeg to convert yuv to mp4. After converting I’m getting mp4 video , but at the end of video there is blurred mess of colours and lines and I got the following error while converting:
@chunsiong.zoom
Again I got the same video with blurred mess of colours and lines at the end and error with respect to frame size.
Also I observed that I’m getting blurred mess of colours and lines only in the part of the video when I start to speak while recording. I tried without speaking and there was no blurred mess of colours and lines in the mp4.
@abhishekmv.dev here’s the tricky part about raw data. the video frames which are returned can be of various size.
I’ve just tried only saving frames which are 720p. If you only choose to save frames which are 720p, there is no corruption in the resulting conversion.
If you want to save all the frames (of various resolution), you will need to do conversion on the fly as you are receiving the yuv420p frames. There are code samples online on how to do it using ffmpeg library.
Hey @abhishekmv.dev as @chunsiong.zoom mentioned, the incoming frames can be different sizes. This causes the errors you observed, as you’re specifying a hardcoded video size with the -video_size 640x360 flag.
Because of this, any frames that are of a different size than 640x360 will be parsed incorrectly and result in an incorrect output frame. This size change can happen dynamically as more videos are subscribed to, or network conditions change.
The most straightforward way to take these incoming frames and turn them into a fixed-resolution output video would be to use GStreamer with an appsrc at the input (so you can specify frame width and height for each frame individually), a videoscale in the middle so you can normalize the frames to your desired output size, and an x264enc + mp4mux to encode and mux the video.
Another alternative is to use Recall.ai for your meeting bots instead. It’s a simple 3rd party API that lets you use meeting bots to get raw audio/video from meetings without you needing to spend months to build, scale and maintain these bots.
