-
-
Notifications
You must be signed in to change notification settings - Fork 115
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Transcript: Queue is stuck when FFmpeg overlay failed #200
Comments
I rolled back my Oryx container version to v5.14.19, and without my PR modify, and there is the same issues. |
Caused by the overlay command failed when driving fix queue: [warn] 2024/07/08 12:00:45.492859 [51][1006] transcript: task uuid=6e61fd0f-194a-49f7-803e-f09d9662028c, live=segments=0, asr=segments=1, fix=segments=3, pat=So, on that note, that minor glitch has been resolved. We have fantastic pitches so far today and a really excellent job from all of the pitchers., overlay=segments=9, config is all=true, key=56B, organization=, base=https://api.openai.com/v1, lang=en, overlay=true, forceStyle=Alignment=2,MarginV=20, videoCodecParams=-c:v h264_nvenc -profile:v main -preset:v medium -bf 0, webvtt=true, webvttCueStyle=STYLE
::cue { background-color: blue; color: FFFFFF; font-family: Helvetica; font-size: 15px; }, webvttCueSetting=line:90% align:center drive fix queue err exit status 1
transcode [-i transcript/355-org-aebe00dc-897f-46d6-9e31-b43c86c9ac3a.ts -vf subtitles=transcript/355-audio-f939a467-5e69-4d1e-b8e2-9a573c85ceed.srt:force_style='Alignment=2,MarginV=20' -c:v h264_nvenc -profile:v main -preset:v medium -bf 0 -c:a aac -copyts -y transcript/355-overlay-47c37aac-5533-44fb-8e8f-43a5c4bace8e.ts]
main.(*TranscriptTask).DriveFixQueue
/g/platform/transcript.go:1998
main.(*TranscriptWorker).Start.func7
/g/platform/transcript.go:1174
runtime.goexit
/usr/local/go/src/runtime/asm_amd64.s:1571 If failed, it will retry forever, stuck the whole queue. |
the failed ffmpeg command log is
|
it seems that if one ffmpeg command failed, then this failed ffmpeg will keep trying again, which will cause the entire queue to appear to not continue running |
There is no problem with streaming with vMix and OBS I found that when AI subtitle stream, such as live/livestream, is pushed back to the same Oryx ai/livestream channel, AI transcription will use the slices of the ai/livestream channel for translation. During the operation, some of the ai/livestream video slices will cause ffmpeg to fail, leading to this problem |
Describe the bug
Use Camera to forward the AI HLS overlay stream, when occur some error, the AI transcription seems to be blocked, and no more transcription result, and the Live Segments Queue became more and more longer
It seems that the original live stream is interrupted for a moment, then pushed the stream again.
Then I suspected that there was an interruption in the original video stream, but I watched the video through Whep and during this period, the video played normally and it seemed to have no interruption
Version
v5.14.22, with this PR #193
To Reproduce
Steps to reproduce the behavior:
transcript
start Transcription
Camera
transcript live stream with subtitle
url as source, and forward the AI stream to third SRT server, SRT server can use Restreamer.Start Live
, run for a while.Expected behavior
AI transcription is running normally
Screenshots
Additional context
The Oryx logs as below, it seems that the source stream is unpublish, but after a while the origin stream is recoverd.
node1. xxxx. com
is the replaced contentOryx.txt
The Oryx_2.txt is another test, and has more logs.
Oryx_2.txt
The text was updated successfully, but these errors were encountered: