srt2hls

Simple audio HLS streaming server.

Intro

The goal of this program is to receive an audio stream with SRT and broadcast it with HLS. It can serve as a CDN origin server or as a set-and-forget streaming server.

How it works

srt2hls use Liquidsoap to receive a stream and encode it in HLS, then it use [nginx] (https://www.nginx.com/) to serve HLS content.

The Liquidsoap container, by default, run radio/live.liq script. It's a fully functionnal example that will :

Receive an SRT input on port 10000
Encode it in aac with 3 quality
Segment it in HLS format in /hls directory

The Nginx container come with a specific configuration to serve HLS content with proper Content-Type, CORS and Cache-Control headers. It need read only access to /hls directory to serve HLS segments and playlists.

Setup

With docker compose

mkdir hls
sudo chown -R 2001:2000 hls
sudo docker-compose up

Local installation requirements

Liquidsoap 1.5.0+ (not released yet, use master)

ffmpeg compiled with fdkaac support (the one on liquidsoap debian/ubuntu repository is fine)

Basic usage

Listening

By default the encoder send blank HLS segments, that mean if it's started, you can already listen the blank stream.

ffplay http://localhost:8080/live.m3u8
vlc http://localhost:8080/live.m3u8

Switch live source

The default script allow to switch the output between inputs.

# Check available input to switch to
curl http://localhost:8080/api/list?livesource
# Switch the livesource
curl http://localhost:8080/api/set?livesource=srt2
curl http://localhost:8080/api/set?livesource=srt1
# check the current live source
curl http://localhost:8080/api/get?livesource

Sending audio to the streaming server

Using ffmpeg

Requirement : ffmpeg compiled with srt support (the one on liquidsoap debian/ubuntu repository is fine)

# static file
ffmpeg -re -i $AUDIOFILE -vn -f wav -codec:a pcm_s16le srt://127.0.0.1:10000
# live stream
ffmpeg -i $LIVESTREAM -vn -f wav -codec:a pcm_s16le srt://127.0.0.1:10000