Live alternative audio with SRT /mpeg-ts

[Daniel_Rossi]

I’m investigating live alternaive language audio with HLS using a custom medialist module. However I’m trying to figure out the workflow for ingest.

The documentation states it needs to ingest a udp mpeg-ts stream. I’m not sure yet how to achieve that with a software encoder yet like gstreamer for testing this. Its easy to probe a VOD file for these generated audio and video PID’s. But I’m confused how it’spossible to identify these tracks for ingest for a live encode ?

Seeing that it’s an mpeg-ts stream, would it be possible to select these PID tracks from an SRT encoder ingest also ?

Right now, I’m investigating the option of two rtmp encodes for the seperate audio also. Would this work ?

[Daniel_Rossi]

Is there anyway to demux audio tracks from an rtmp published stream ? Like the VOD module I suppose ? It seems mpeg-ts is unusuitable as an option as is somehow publishing two rtmp streams.

[Daniel_Rossi]

If I setup an mpeg-ts stream target for an RTMP stream back to the server host, would this be a possible hack to ingest to the multiple audio tracks ?

I’m investigating this option now, so publish with two audio tracks,but I would have no idea about PID’s. Mpeg-ts encoders are not easily accessible for end users, I can’t seem to find suitable software encoders for this setup either ?

I’m trying to find ways to publish multi channel mpeg-ts with software as a test also but no luck.

[Rose_Power-Wowza_Com]

We’re not sure since @Daniel Rossi since we have not tested this workflow to the point we officially support it, but would be happy to take a closer look through a support ticket and test it out with you.

https://www.wowza.com/support/open-ticket

[Daniel_Rossi]

PID’s are being delivered over SRT mpeg-ts. I am trying to work on some tests. But trying to get UDP mpeg-ts working first which I’ve successfully done via OBS.

[kv_pham]

FYI there is already this suggestion for srt: /users/Daniel_Rossi

[Daniel_Rossi]

I was able to properly run a encoder test using ffmpeg as gstreamer has bugs. audio PID’s are sent with the mpegts

ffmpeg -fflags +genpts -re -i sintel_lang_2000k.mp4 -acodec copy -vcodec copy -map 0:v:0 -map 0:a:0 -map 0:a:2 -strict -2 -y -f mpegts ‘srt://192.168.4.43:8081?pkt_size=1316&mode=caller’

Audio languages can be ingested uring SRT, after the stream file is modified

{ uri : “srt://0.0.0.0:8081”, mpegtsVideoPID : “0”, mpegtsAudioPID : “258” }

My custom media list generates a rendition for the language stream file and the one with both video and default language audio.

Works like UDP mpegts.

[Daniel_Rossi]

It was working now nothing. Only one of the ports opens !

[mark_rump]

Did you make it work? I have the same issue … UDP port sharing not working with SRT. Restreaming the SRT via UDP to the same instance loses my additional 2nd and 3rd audio tracks.

[Nathan_Ihlenfeldt]

I did.I ran a ffmpeg docker container on the WSE server to convert SRT to MPEG-TS UDP. FFMpeg handles the SRT negotiation, and you’ll create udp:// stream files in WSE. This is not possible with WSE alone at this time