FFmpeg cookbook

Here are some examples of what is possible to do with the ffmpeg support in liquidsoap:

Relaying without re-encoding

With ffmpeg support, Liquidsoap can relay encoded streams without re-encoding them, making it possible to re-send a stream to multiple destinations. Here’s an example:

# Input the stream,
# from an Icecast server or any other source
encoded_source = input.http("https://icecast.radiofrance.fr/fip-hifi.aac")

# Send to one server here:
output.icecast(
  %ffmpeg(format="adts", %audio.copy),
  fallible=true,
  mount="/restream",
  host="streaming.example.com", port=8000, password="xxx",
  encoded_source)

# An another one here:
output.icecast(
  %ffmpeg(format="adts", %audio.copy),
  fallible=true,
  mount="/restream",
  host="streaming2.example.com", port=8000, password="xxx",
  encoded_source)

We cannot use mksafe here because the content is not plain pcm samples, which this operator is designed to produce. There are several ways to make the source infallible, however, either by providing a single(...) source with the same encoded content as we expect from encoded_source or by creating an infallible source using ffmpeg.encode.audio.

On-demand relaying without re-encoding

Another refinement on the previous example is the capacity to relay a stream only when listeners are connected to it, all without re-encoding the content.

To make it work, you will need a format that can be handled by ffmpeg for that purpose. mp3 is a good example.

In the script below, you need to match the encoded format of the stream with a blank file (or any other file). The output.harbor will then relay the data from the file if no one is connected and start/stop the underlying input when there are listeners:

stream = input.http(start = false, "https://wwoz-sc.streamguys1.com/wwoz-hi.mp3")

listeners_count = ref(0)

def on_connect(~headers, ~uri, ~protocol, _) =
  listeners_count := !listeners_count + 1
  if !listeners_count > 0 and not stream.is_started() then
    log("Starting input")
    stream.start()
  end
end

def on_disconnect(_) =
  listeners_count := !listeners_count - 1
  if !listeners_count == 0 and stream.is_started() then
    log("Stopping input")
    stream.stop()
  end
end

blank = single("/tmp/blank.mp3")

stream = fallback(track_sensitive=false, [stream, blank])

output.harbor(
  %ffmpeg(format="mp3", %audio.copy),
  format="audio/mpeg",
  mount="relay",
  on_connect=on_connect,
  on_disconnect=on_disconnect,
  stream)

Shared encoding

Liquidsoap can also encode in one place and share the encoded with data with multiple outputs, making it possible to minimize CPU resources. Here’s an example adapted from the previous one:

# Input the stream,
# from an Icecast server or any other source
source = input.http("https://icecast.radiofrance.fr/fip-hifi.aac")

# Make it infallible:
source = mksafe(source)

# Encode it in mp3:
source = ffmpeg.encode.audio(
  %ffmpeg(%audio(codec="libmp3lame")),
  source)

# Send to one server here:
output.icecast(
  %ffmpeg(format="mp3", %audio.copy),
  mount="/restream",
  host="streaming.example.com", port=8000, password="xxx",
  source)

# An another one here:
output.icecast(
  %ffmpeg(format="mp3", %audio.copy),
  mount="/restream",
  host="streaming2.example.com", port=8000, password="xxx",
  source)

Shared encoding is even more useful when dealing with video encoding, which is very costly. Here’s a fun example sharing audio and video encoding and sending to different destinations, both via Icecast and to YouTube/Facebook via the rtmp protocol:

# An audio source
audio = ...

# Encode it in mp3
audio = ffmpeg.encode.audio(
  %ffmpeg(%audio(codec="libmp3lame")),
  audio)

# Send it to icecast
output.icecast(
  %ffmpeg(format="mp3", %audio.copy),
  host = "...",
  password = "...",
  mount = "/stream",
  audio
)

# A video source, for instance a static image
video = ...

# Encode it in h264 format
video = ffmpeg.encode.video(
  %ffmpeg(%video(codec="libx264")),
  video)

# Mux it with the audio
stream = source.mux.video(video=video, audio)

# Copy encoder for the rtmp stream
enc = %ffmpeg(
  format="flv",
  %audio.copy,
  %video.copy)

# Send to YouTube
key = "..."
url = "rtmp://a.rtmp.youtube.com/live2/#{key}"
output.url(url=url, enc, stream)

# Send to Facebook
key = "..."
url = "rtmps://live-api-s.facebook.com:443/rtmp/#{key}"
output.url(self_sync=true, url=url, enc, stream)

Add transparent logo and video

See: https://github.com/savonet/liquidsoap/discussions/1862

Live switch between encoded content

This is an ongoing development effort. Please refer to the online support channels if you are experiencing issues with this kind of feature.

Starting with liquidsoap 2.1.x, it is gradually becoming possible to do proper live switches on encoded content and send the result to different outputs.

Please note that this requires a solid knowledge of media codecs, containers and ffmpeg bitstream filters. Different input and output containers store codec binary data in different ways and those are not always compatible. This requires the use of bitstream filters to adapt the binary data and, it’s possible some new filters will need to be written to support more combinations of input/output and codecs.

Here’s a use case that has been tested: live switch between a playlist of mp4 files and a rtmp flv input:

s1 = input.rtmp(listen=false,"rtmp://....")
s1 = ffmpeg.filter.bitstream.h264_mp4toannexb(s1)

s2 = playlist("/path/to/playlist")
s2 = ffmpeg.filter.bitstream.h264_mp4toannexb(s2)

s = fallback(track_sensitive=false, [s1, s2])

mpegts = %ffmpeg(
  format="mpegts",
  fflags="-autobsf",
  %audio.copy, %video.copy)

streams = [
  ("mpegts",mpegts),
]

output_dir = "/tmp/hls"

output.file.hls(playlist="live.m3u8",
                fallible=true,
                segment_duration=5.,
                output_dir,
                streams,
                s)
  • We need the h264_mp4toannexb on each stream to make sure that the mp4 data conforms to what the mpegts container expect
  • We need to disable ffmpeg’s automatic insertion of bitstream filters via -autobsf. FFmpeg does not support this kind of live switch at the moment and its automatically inserted filters won’t work, which is why we’re doing it ourselves.

That’s it! In the future we want to extend this use-case to also be able to output to a rtmp output from the same data. And more!