On Wed, Jan 30, 2019 at 5:09 PM Gilles Pietri <contact+...@gilouweb.com> wrote:
>
> Le 30/01/2019 à 12:53, Damian a écrit :
> > Hi,
> >
> > I am running liquidsoap 1.3.6 and I have followed the instructions for
> > applying replay_gain in the docs here
> > https://www.liquidsoap.info/doc-1.3.3/replay_gain.html
> > I believe that I have added the code for replay_gain correctly in my
> > short script below, but volume is uneven across tracks in my stream.
>
In this script, replay gain works as expected, though it's read from
an annotated file.
The playlist file looks like this:
...
annotate:liq_cue_in="0.000",liq_start_next="5.000",duration="463.800",liq_amplify="5.300dB":/home/john/src/radio/mez/04
Vildgolia (Deaf, Dumb & Blind).NpsjiF.mka
annotate:liq_cue_in="0.000",liq_start_next="4.050",duration="305.950",liq_amplify="-1.200dB":/home/john/src/radio/mez/13
- Mr. Blue Sky.feOPPs.mka
...
And here's the script.
#!~/.opam/system/bin/liquidsoap
# Make a log
set("scheduler.fast_queues", 2)
set("scheduler.generic_queues", 5)
set("scheduler.non_blocking_queues", 4)
#set("scheduler.log", true)
set("log.file.path", "/tmp/basic-radio.log")
set("server.telnet", true)
set("server.telnet.bind_addr", "127.0.0.1")
set("server.telnet.port", 1234)
set("server.timeout", -1.0)
set("decoder.file_decoders",
["META","WAV","AIFF","MIDI","IMAGE","FFMPEG","FLAC","AAC","MP4","OGG","MAD"])
set("decoder.file_extensions.gstreamer", [])
set("decoder.mime_types.gstreamer", [])
set("decoder.file_extensions.ffmpeg",["mp3","mp4","m4a","wav","flac","ogg","webm","opus","mka"])
set("protocol.external.curl", "/usr/bin/curl")
set("frame.audio.samplerate", 32000)
set("request.metadata_decoders.duration", false)
#set("ffmpeg.log.verbosity", "verbose")
# My function to break for news. Copied from
# fallback.skip, but made track sensitive
# THIS REALLY DOESN'T WORK YET. The
# incoming audio (from the playlist) doesn't fade out.
# But the fade-in of the next track works. Why?
def break.news(~input,f)
def transition1(a,b) =
b
end
def transition2(a,b) =
add ([fade.initial(duration=0.01,type="exp",b), a])
end
fallback(track_sensitive=true,transitions=[transition1,transition2],[input,f])
end
security = single("/home/john/src/radio/fault.flac")
# Here's the playlist, cut as per our Python pre-production program dictates
myplaylist = amplify(override="liq_amplify", 1.0,
cue_cut(playlist(length=60.0,
"/home/john/src/radio/test-nodup-shuf-processed.m3u8")))
# Now do the crossfades. Each track's end is determined by its fade-out
# given by the Python program, so they can overlap without clashing too much.
#
myplaylist = break.news(input=request.queue(id="override"), myplaylist)
myplaylist = crossfade(fade_out=0.01, fade_in=0.01, conservative=true,
myplaylist)
# Make the playlist secure
radio = fallback(track_sensitive = false, [myplaylist, security])
# Here, we use FFmpeg to process the audio, and convert it ready for
transmission.
# It does not exit cleanly, for reasons I don't understand yet.
#
# We could use output.external, but metadata (e.g. what's playing)
doesn't get transferred that way.
# The FFmpeg line is like a broadcast audio chain: we normalize the
loudness over a few seconds
# then multi-band compress it, then add a single-band limiter before
EBU R.128 normalization with a
# target of -14LUFS.
# The AAC stream is at about 38 kbit/s (variable), so should work
almost everywhere.
output.icecast(description="Experimental stream using Liquidsoap",
genre="Freeform", name="Music Too", host="warblefly.sytes.net",
port=8000, mount="audio.aac", public=true,
url="http://warblefly.sytes.net:8000/audio.aac";, format="audio/aac",
password="", %external(samplerate=32000, channels=2, process="ffmpeg
-err_detect ignore_err -report -f s16le -ar 32000 -ac 2 -i pipe:0
-acodec libfdk_aac -vbr 3 -profile:a aac_he_v2 -vn -af
dynaudnorm=g=31:m=70:r=1.0,asetnsamples=2048,volume=-18dB,mcompand='0.005\,0.1
6 -47/-40\,-34/-34\,-17/-33 100 | 0.003\,0.05 6
-47/-40\,-34/-34\,-17/-33 400 | 0.000625\,0.0125 6
-47/-40\,-34/-34\,-15/-33 1600 | 0.0001\,0.025 6
-47/-40\,-34/-34\,-31/-31\,-0/-30 6400 | 0\,0.025 6
-38/-31\,-28/-28\,-0/-25
15999',volume=+20dB,alimiter=-4dB,loudnorm=I=-14:LRA=1.0:TP=-0.25:linear=0,aresample=32000
-f adts pipe:1"), radio)
_______________________________________________
Savonet-users mailing list
Savonet-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/savonet-users
