HLS-24: Encode fails where audio stream present but null

Issue Information

Issue Type: Bug
 
Priority: Major
Status: Closed

Reported By:
Ben Tasker
Assigned To:
Ben Tasker
Project: HLS Stream Creator (HLS)
Resolution: Done (2019-07-26 13:17:37)
Affects Version: 1.0,
Target version: 1.0,
Components: Transcoding ,
Labels: Audio,

Created: 2017-07-16 14:55:46
Time Spent Working
Estimated:
 
45 minutes
Remaining:
 
45 minutes
Logged:
 
0 minutes


Description
HLS Stream creator will fail to run where an audio stream is present but shows as type none 
  Duration: 00:00:00.00, start: 4241.278000, bitrate: N/A
    Stream #0:0: Video: flv1, 1 reference frame, yuv420p, 1280x720, 199 kb/s, 8.33 tbr, 1k tbn, 1k tbc
    Stream #0:1: Audio: none, 0 channels


The resulting error is 
  Duration: 00:00:00.00, start: 4241.278000, bitrate: N/A
    Stream #0:0: Video: flv1, 1 reference frame, yuv420p, 1280x720, 199 kb/s, 8.33 tbr, 1k tbn, 1k tbc
    Stream #0:1: Audio: none, 0 channels
[graph 0 input from stream 0:0 @ 0xace980] w:1280 h:720 pixfmt:yuv420p tb:1/1000 fr:25/3 sar:0/1 sws_param:flags=2
No decoder for stream #0:1, filtering impossible
Error opening filters!
Invoking deleteStream


This can be mitigated by setting an FFMPEG flag to tell ffmpeg to ignore the audio stream: 
ben@milleniumfalcon:/tmp/HLSCreate$ FFMPEG_FLAGS="-an"
ben@milleniumfalcon:/tmp/HLSCreate$ export FFMPEG_FLAGS
ben@milleniumfalcon:/tmp/HLSCreate$ ./HLS-Stream-Creator.sh -l -i rtmp://ent1/camtest/camera -s 2


But it'd be nice to have a commandline flag to be able to do this


Issue Links

relates to HLS-25: Additional flags/features to aid Low Latency Streaming
Toggle State Changes

Activity


2019-07-26 13:09:25

Repo: HLS-Stream-Creator
Commit: db85e45814b2a163c5ae927bca36a74460483fe7
Author: B Tasker <github@<Domain Hidden>>

Date: Fri Jul 26 13:03:02 2019 +0100
Commit Message: HLS-24 Allow audio track to be ignored with the -n commandline argument

This forces -an to be passed to ffmpeg.

It's not generally required, as missing audio tracks will be handled by default, but this allows the user to address inputs where `ffmpeg` reports the Audio type as `none`: 
    Stream #0:0: Video: flv1, 1 reference frame, yuv420p, 1280x720, 199 kb/s, 8.33 tbr, 1k tbn, 1k tbc
    Stream #0:1: Audio: none, 0 channels

This test stream was the result of a webcam being published into an RTMP relay with FFMPEG and then pulled back out the other side.



Modified (-)(+)
-------
HLS-Stream-Creator.sh
README.md




Webhook User-Agent 

GitHub-Hookshot/20c7c42


View Commit


2019-07-26 13:17:17
Commit db85e4 implements a new commandline flag -n

When this flag is present in the HLS-Stream-Creator commandline, we'll pass ffmpeg the flag -an in order to disable audio.

This leads to a slightly odd (but fine) ffmpeg command, in that we specify the audio codec to use, and then disable audio anyway. 
ffmpeg -i ~/How_to_Build_a_Cathedral--.mp4 -y -vcodec libx264 -acodec aac -threads 0 -map 0 -flags -global_header -f segment -segment_list ./output/How_to_Build_a_Cathedral--.mp4.m3u8 -segment_time 10 -segment_format mpeg_ts -an ./output/How_to_Build_a_Cathedral--.mp4_%05d.ts

It's a little odd looking but doesn't cause any harm, so it seems foolish to introduce additional complexity to remove arguments from a command the user will never routinely see anyway.

2019-07-26 13:17:30
btasker added 'Audio' to labels

2019-07-26 13:17:37
btasker changed status from 'Open' to 'Resolved'

2019-07-26 13:17:37
btasker added 'Done' to resolution

2019-07-26 13:17:41
btasker changed status from 'Resolved' to 'Closed'