Liquidsoap 2.3.1 (rolling release) : USB Alsa Input - Icecast Output - intermittent and juddery output (clock issues? - buffer overrun) #4236

mojikosu · 2024-12-02T18:01:41Z

mojikosu
Dec 2, 2024

Dear Community,
I am running a Raspberry Pi 3+ on Raspbian Bookworm 32-bit, with the newly released Liquidsoap 2.3.0.
I complie liquidsoap sucessfully via opam, pinning opam to the up-to-date github repository. ffmpeg is compiled from source and integrated into the liquidsoap build (for HLS output).

As I did previously under liquidsoap 2.2.5, I capture sound via a USB audio interface through ALSA and output the stream via icecast encoded as MP3 stream (lame encoder) and AAC HLS output.

Issue
Whereas there were absolutely no issues with that routing, under liquidsoap 2.3.0 the sound output (icecast and HLS) is all juddery and cuts off every 5 seconds for a duration of around 2 seconds.

Could this be a clock or buffer issue? Who has an idea how to address this?

Any kind of help is highly appreciated.

Thanks in advance, dear community

mojikosu

Below is the relevant part of the liq file (only mp3 icecast output)

#set the path and permissions for the logfile
set("log.file.path","/var/log/liquidsoap/VinylfromWax.log")
set("log.file.perms",777)
set("log.unix_timestamps",true)
#==============input from soundcard

s = (input.alsa(device = "pcm.default"))
s = amplify(2.2 ,s)
s = bass_boost(frequency=140.0 ,gain=1.2 ,s)
#======================FORMATS======================================================
#-----------------MP3 Stream with floating point encoder
f = %mp3.cbr(samplerate=44100,bitrate=320)
#===================================================================================

#======Prepare source to accept metadata injection=========
s = insert_metadata(s)
#================ Metadata grab and update section =========================================
#======== Launch songrec to detect one song, output as JSON, insert metadata into source stream========
def metadataupdater()
  json =  process.read("/usr/sbin/songrec recognize --json")
  let json.parse ({
    track = {
      subtitle,
      title,
      images = {
        coverart }
     }
    } : {
     track : {
      subtitle : string,
      title : string,
      images : {
       coverart: string }
    }
    }) = json

  s.insert_metadata([("title", title),("artist", subtitle),("coverart", coverart)])
end

#This intermediate function runs the actual metadata updater in a thread, so that the
#metadata update does *not* interrupt the stream.
def handleblank()
 thread.run(metadataupdater)
end

silent = ref(false)

s = blank.detect(
  threshold=-21.,
  max_blank=1.,
  on_noise=handleblank,
  {silent := true},
  s)

#==============CREATE HTTP SERVER FOR METADATA OUTPUT================
meta = ref([])

# s = some source
s.on_metadata(fun (m) -> meta := m)
# Return the json content of meta
def get_meta(_, response) =
  response.json(meta())
end

# Register get_meta at port 700
harbor.http.register(port=7000,method="GET","/getmeta",get_meta)

#=================OUTPUT==================
output.icecast(f,
host = "127.0.0.1",
port = 8000,
password = "WaxVinyl",
mount = "vinyl",
name = "Vinyl from Wax",
id="Vinyl Station",
on_start=handleblank,
send_icy_metadata=true,
description="Vinyl from Wax",
#url="http://IP-ADDRESS&#8221;,
s)

vitoyucepi · 2024-12-03T06:07:49Z

vitoyucepi
Dec 3, 2024
Collaborator

Hi @mojikosu,
I don't know exactly what the problem is. Maybe it's caused by #3806.
Try to use buffer(input.alsa()).

@toots, I didn't see any notes about input.alsa in the migration guide.

6 replies

vitoyucepi Dec 3, 2024
Collaborator

I think you can monitor the buffer size if you increase its maximum size.

s = buffer(input.alsa(device = "pcm.default"), buffer=5., max=30.)
thread.run(
  fun() -> begin
    a = float(s.buffer_length())
    b = float(settings.frame.audio.samplerate())
    print("#{a} #{b} #{a/b}")
  end,
  every=1.
)

s = mksafe(s)

mojikosu Dec 3, 2024
Author

@vitoyucepi
Hi again,
I implemented your suggestions. And thanks again for your assistance!
The buffer depletes eventually, please see the attached terminal output.
It's so weird, because I had no issues like that under liquidsoap 2.2.5.
Any clue on this? The hardware setup has not changed at all.

$ liquidsoap /etc/liquidsoap/vinylfromWax.liq 
0.0 44100.0 0.0
36162.0 44100.0 0.82
75852.0 44100.0 1.72
109368.0 44100.0 2.48
153468.0 44100.0 3.48
193158.0 44100.0 4.38
221382.0 44100.0 5.02
215208.0 44100.0 4.88
208152.0 44100.0 4.72
208152.0 44100.0 4.72
209034.0 44100.0 4.74
207270.0 44100.0 4.7
205506.0 44100.0 4.66
199332.0 44100.0 4.52
196686.0 44100.0 4.46
196686.0 44100.0 4.46
194922.0 44100.0 4.42
179928.0 44100.0 4.08
181692.0 44100.0 4.12
171990.0 44100.0 3.9
168462.0 44100.0 3.82
166698.0 44100.0 3.78
161406.0 44100.0 3.66
160524.0 44100.0 3.64
156996.0 44100.0 3.56
153468.0 44100.0 3.48
152586.0 44100.0 3.46
150822.0 44100.0 3.42
143766.0 44100.0 3.26
138474.0 44100.0 3.14
136710.0 44100.0 3.1
129654.0 44100.0 2.94
126126.0 44100.0 2.86
123480.0 44100.0 2.8
111132.0 44100.0 2.52
111132.0 44100.0 2.52
110250.0 44100.0 2.5
104958.0 44100.0 2.38
92610.0 44100.0 2.1
85554.0 44100.0 1.94
82908.0 44100.0 1.88
76734.0 44100.0 1.74
77616.0 44100.0 1.76
69678.0 44100.0 1.58
62622.0 44100.0 1.42
64386.0 44100.0 1.46
59976.0 44100.0 1.36
51156.0 44100.0 1.16
48510.0 44100.0 1.1
44982.0 44100.0 1.02
35280.0 44100.0 0.8
28224.0 44100.0 0.64
30870.0 44100.0 0.7
21168.0 44100.0 0.48
14112.0 44100.0 0.32
0.0 44100.0 0.0 #<=== this is where the stream goes to blank and resumes a few seconds later
34398.0 44100.0 0.78
72324.0 44100.0 1.64 ```

vitoyucepi Dec 4, 2024
Collaborator

On average, the ALSA interface produced about 39000 samples per second. This is 88% of the required speed.

mojikosu Dec 4, 2024
Author

@vitoyucepi
Thanks for your insight.
This pointed me into the right direction.
I rewrote the alsa config files I used previously:
asound.conf specified a specific buffer and period size and the sampling rate at 44100. I removed these lines and as far as I have tested now the issue seems to have gone.
Thank you again for your time and assistance :)

mojikosu Dec 4, 2024
Author

for reference, this is the now working asound.conf:

pcm.!default {
    type plug
    slave.pcm "liquidsoap"
}

pcm.liquidsoap {
        type dsnoop
        ipc_key 5978293 # must be unique for all dmix plugins!!!!
        ipc_key_add_uid yes
        slave {
                pcm "hw:0,0"
                channels 2
#                period_size 1024
#               buffer_size 4096
#               rate 44100
#               periods 0 
#               period_time 0
        }
        bindings {
                0 0
                1 1
        }
}

mojikosu · 2024-12-08T18:45:28Z

mojikosu
Dec 8, 2024
Author

Hi again,

Thanks @toots and @vitoyucepi for your code updates and support in relation to this issue so far.
Unfortunately, the buffer issues under Liquidsoap 2.3.1 (rolling release) and ALSA persist. (Config under Liquidsoap 2.2.5 ran fine.)
I have implemented buffer_size for input.alsa() but I am yet to see results.

I must be getting something very wrong here. Any kind of help is so much welcome.

My current ALSA configuration per asound.conf:

pcm.!default {
    type plug
    slave.pcm "liquidsoap"
}

pcm.liquidsoap {
    type dsnoop
    ipc_key 5978293 # Unique IPC key
    ipc_key_add_uid yes
    slave {
        pcm "hw:1,0"
        channels 2
        rate 44100        # Set sample rate to 44.1 kHz
        period_size 1024  # Frames per period (low latency: 23.2 ms)
        buffer_size 4096  # Total buffer size (92.9 ms total latency)
    }
    bindings {
        0 0
        1 1
    }
}

Here is my .liq file

# set the path and permissions for the logfile
settings.log.file.path :=　"/var/log/liquidsoap/VinylfromWax.log"
settings.log.file.perms := 777
settings.log.unix_timestamps := false
#==============ALSA Settings==========
#settings.alsa.alsa_buffer := 0
#settings.alsa.periods := 0

#==============Input from soundcard
s = (input.alsa(device = "pcm.default", self_sync=true, buffer_size=0.05):source(audio=pcm(stereo)))
s = amplify(2.2 ,s)
s = bass_boost(frequency=140.0 ,gain=1.2 ,s)
#======================FORMATS======================================================
#-----------------MP3 Stream with floating point encoder
f = %mp3.cbr(samplerate=44100,bitrate=320)

#=================OUTPUT==================
output.icecast(f,
host = "127.0.0.1",
port = 8000,
password = "WaxVinyl",
mount = "vinyl",
name = "Vinyl from Wax",
id="Vinyl Station",
on_start=handleblank,
send_icy_metadata=true,
s)

And here's the log:

2024/12/08 19:37:28 >>> LOG START
2024/12/08 19:37:25 [main:3] Liquidsoap rolling-release-v2.3.x-1-g83173fa+dev
2024/12/08 19:37:25 [main:3] Using: alsa=0.3.0 angstrom=0.16.1 base=v0.16.3 base.base_internalhash_types=v0.16.3 base.caml=v0.16.3 base.shadow_stdlib=v0.16.3 bigarray=[distributed with Ocaml] bigstringaf=0.10.0 bytes=[distributed with OCaml 4.02 or above] camlp-streams camomile.lib=2.0 cry=1.0.3 curl=0.9.2 domain_shims dtools=0.4.5 dune-build-info=3.17.0 dune-private-libs.dune-section=3.17.0 dune-site=3.17.0 dune-site.private=3.17.0 duppy=0.9.4 fileutils=0.6.4 flac=0.5.0 flac.decoder=0.5.0 flac.ogg=0.5.0 gen=1.1 lame=0.3.7 liquidsoap-lang=rolling-release-v2.3.x-1-g83173fa liquidsoap-lang.console=rolling-release-v2.3.x-1-g83173fa liquidsoap_alsa=rolling-release-v2.3.x-1-g83173fa-dirty liquidsoap_builtins=rolling-release-v2.3.x-1-g83173fa-dirty liquidsoap_core=rolling-release-v2.3.x-1-g83173fa-dirty liquidsoap_flac=rolling-release-v2.3.x-1-g83173fa-dirty liquidsoap_lame=rolling-release-v2.3.x-1-g83173fa-dirty liquidsoap_mad=rolling-release-v2.3.x-1-g83173fa-dirty liquidsoap_ogg=rolling-release-v2.3.x-1-g83173fa-dirty liquidsoap_ogg_flac=rolling-release-v2.3.x-1-g83173fa-dirty liquidsoap_optionals=rolling-release-v2.3.x-1-g83173fa-dirty liquidsoap_runtime=rolling-release-v2.3.x-1-g83173fa-dirty liquidsoap_shine=rolling-release-v2.3.x-1-g83173fa-dirty liquidsoap_vorbis=rolling-release-v2.3.x-1-g83173fa-dirty mad=0.5.3 magic-mime=1.3.1 mem_usage=0.1.1 menhirLib=20240715 metadata=0.3.0 mm=0.8.6 mm.audio=0.8.6 mm.base=0.8.6 mm.image=0.8.6 mm.midi=0.8.6 mm.video=0.8.6 ogg=0.7.4 ogg.decoder=0.7.4 ppx_compare.runtime-lib=v0.16.0 ppx_hash.runtime-lib=v0.16.0 ppx_sexp_conv.runtime-lib=v0.16.0 re=1.12.0 saturn_lockfree=0.4.1 sedlex=3.3 seq=[distributed with OCaml 4.07 or above] sexplib0=v0.16.0 shine=0.2.3 stdlib-shims=0.3.0 str=[distributed with Ocaml] stringext=1.6.0 threads=[distributed with Ocaml] threads.posix=[internal] unix=[distributed with Ocaml] uri=4.4.0 vorbis=0.8.0 vorbis.decoder=0.8.0
2024/12/08 19:37:25 [clock:3] Using builtin (low-precision) implementation for latency control
2024/12/08 19:37:28 [main:3] User script loaded in 2.84 seconds.
2024/12/08 19:37:28 [frame:3] Using 44100Hz audio, 25Hz video, 44100Hz main.
2024/12/08 19:37:28 [frame:3] Video frame size set to: 1280x720
2024/12/08 19:37:28 [frame:3] Targeting 'frame.duration': 0.02s = 882 ticks.
2024/12/08 19:37:28 [sandbox:3] Sandboxing disabled
2024/12/08 19:37:28 [startup:3] main script hash computation: 0.23s
2024/12/08 19:37:28 [startup:3] Loading main script from cache!
2024/12/08 19:37:28 [startup:3] main script cache retrieval: 0.36s
2024/12/08 19:37:28 [startup:3] Evaluating main script: 0.19s
2024/12/08 19:37:28 [clock:3] Starting clock generic with 11 source(s) and sync: auto
2024/12/08 19:37:28 [harbor:3] Adding handler for 'GET ^/getmeta$' on port 7000
2024/12/08 19:37:28 [video.converter:3] Couldn't find preferred video converter: ffmpeg.
2024/12/08 19:37:28 [audio.converter:3] Using samplerate converter: native.
2024/12/08 19:37:28 [alsa_in(pcm.default):3] Content type is {audio=pcm(stereo)}.
2024/12/08 19:37:28 [alsa_in(pcm.default):3] Using ALSA 1.2.4.
2024/12/08 19:37:28 [alsa_in(pcm.default):3] Could not set buffer size to: 0.05s (2205 samples), got: 0.05 (2048 samples).
2024/12/08 19:37:28 [alsa_in(pcm.default):3] Samplefreq=44100Hz, Bufsize=2048B, Frame=4B, Periods=2
2024/12/08 19:37:28 [track_amplify:3] Content type is {audio=pcm(stereo)}.
2024/12/08 19:37:28 [amplify:3] Content type is {audio=pcm(stereo)}.
2024/12/08 19:37:28 [biquad_filter:3] Content type is {audio=pcm(stereo)}.
2024/12/08 19:37:28 [compress:3] Content type is {audio=pcm(stereo)}.
2024/12/08 19:37:28 [source.2:3] Content type is {audio=pcm(stereo)}.
2024/12/08 19:37:28 [audio.add:3] Content type is {audio=pcm(stereo)}.
2024/12/08 19:37:28 [source.3:3] Content type is {audio=pcm(stereo)}.
2024/12/08 19:37:28 [insert_metadata:3] Content type is {audio=pcm(stereo)}.
2024/12/08 19:37:28 [blank.detect:3] Content type is {audio=pcm(stereo)}.
2024/12/08 19:37:28 [Vinyl_Station:3] Content type is {audio=pcm(stereo)}.
2024/12/08 19:37:28 [Vinyl_Station:3] Content type is {audio=pcm(stereo)}.
2024/12/08 19:37:28 [video.text:3] Using native implementation
2024/12/08 19:37:28 [Vinyl_Station:3] Connecting mount vinyl for [email protected]...
2024/12/08 19:37:28 [Vinyl_Station:3] Connection setup was successful.
2024/12/08 19:37:28 [alsa_in(pcm.default):2] Overrun! You may minimize them by increasing the buffer size.
2024/12/08 19:37:28 [alsa_in(pcm.default):2] Trying to recover..
2024/12/08 19:37:28 [alsa_in(pcm.default):2] Overrun! You may minimize them by increasing the buffer size.
2024/12/08 19:37:28 [alsa_in(pcm.default):2] Trying to recover..
2024/12/08 19:37:28 [alsa_in(pcm.default):2] Overrun! You may minimize them by increasing the buffer size.
2024/12/08 19:37:28 [alsa_in(pcm.default):2] Trying to recover..
2024/12/08 19:37:28 [alsa_in(pcm.default):2] Overrun! You may minimize them by increasing the buffer size.
2024/12/08 19:37:28 [alsa_in(pcm.default):2] Trying to recover..
2024/12/08 19:37:29 [alsa_in(pcm.default):2] Overrun! You may minimize them by increasing the buffer size.
2024/12/08 19:37:29 [alsa_in(pcm.default):2] Trying to recover..
2024/12/08 19:37:29 [alsa_in(pcm.default):2] Overrun! You may minimize them by increasing the buffer size.
2024/12/08 19:37:29 [alsa_in(pcm.default):2] Trying to recover..
2024/12/08 19:37:29 [alsa_in(pcm.default):2] Overrun! You may minimize them by increasing the buffer size.
2024/12/08 19:37:29 [alsa_in(pcm.default):2] Trying to recover..
2024/12/08 19:37:29 [alsa_in(pcm.default):2] Overrun! You may minimize them by increasing the buffer size.
2024/12/08 19:37:29 [alsa_in(pcm.default):2] Trying to recover..
2024/12/08 19:37:29 [alsa_in(pcm.default):2] Overrun! You may minimize them by increasing the buffer size.
2024/12/08 19:37:29 [alsa_in(pcm.default):2] Trying to recover..
2024/12/08 19:37:29 [alsa_in(pcm.default):2] Overrun! You may minimize them by increasing the buffer size.
2024/12/08 19:37:29 [alsa_in(pcm.default):2] Trying to recover..
2024/12/08 19:37:29 [alsa_in(pcm.default):2] Overrun! You may minimize them by increasing the buffer size.
2024/12/08 19:37:29 [alsa_in(pcm.default):2] Trying to recover..

and so it continues...

8 replies

mojikosu Dec 10, 2024
Author

@toots

Hi again,
as you suggested, I have now installed JACK and recompiled liquidsoap from opam with bjack enabled.

I start the jack1 server with this command line:

jackd -n default -R -d alsa -S -D hw:0,0 -r 44100 -p 1024 -n3

Which provides this output:

jackd 0.126.0
Copyright 2001-2009 Paul Davis, Stephane Letz, Jack O'Quinn, Torben Hohn and others.
jackd comes with ABSOLUTELY NO WARRANTY
This is free software, and you are welcome to redistribute it
under certain conditions; see the file COPYING for details

JACK compiled with System V SHM support.
loading driver ..
apparent rate = 44100
creating alsa driver ... hw:0|hw:0|1024|3|44100|0|0|nomon|swmeter|-|16bit
configuring for 44100Hz, period = 1024 frames (23.2 ms), buffer = 3 periods
ALSA: final selected sample format for capture: 16bit little-endian
ALSA: use 3 periods for capture
ALSA: final selected sample format for playback: 16bit little-endian
ALSA: use 3 periods for playback

I have adapted the corresponding .liq file to read

s = input.jack(server="default")

Liquidsoap starts up without issues and broadcast its stream through icecast.
However, there is no sound in the icecast stream.

The Alsa driver that JACK creates, on the other hand, delivers sound, though.
see /etc/asound.conf:

pcm.!default {
    type plug
    slave.pcm "jack"
}

The followingJACK server files are created in :

/dev/shm/jack-1000/default $ 

__db.001  __db.002  __db.003  jack_0  jack_ack_0  jack-ack-fifo-5935-0  jack-ack-fifo-5935-1  jack-ack-fifo-5935-2  metadata.db

So that tells me the server is up, and that liquidsoap is connecting to it.

Do you spot, where I am missing something?

Many thanks in advance for your efforts and patience.

toots Dec 10, 2024
Maintainer

You probably need to connect the alsa input to liquidsoap in jack. The GUI for that is called qjackctl.

mojikosu Dec 10, 2024
Author

As I run everything on a headless Raspberry Pi, I'll see if there is a way of doing this via command line.
Or do you happen to know the right command right away? ;)
Thanks again for your time :)

toots Dec 11, 2024
Maintainer

I think jack has some headless methods but I'm not familiar with them. The other solution is to increase the ALSA buffer until the problem goes away.. 🙂

mojikosu Dec 11, 2024
Author

I hit a wall with jack, currently, and I will move to testing with pulseaudio today. :)

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Liquidsoap 2.3.1 (rolling release) : USB Alsa Input - Icecast Output - intermittent and juddery output (clock issues? - buffer overrun) #4236

{{title}}

{{editor}}'s edit

{{editor}}'s edit

Replies: 2 comments 14 replies

{{title}}

{{title}}

{{title}}

{{editor}}'s edit

{{editor}}'s edit

{{title}}

{{title}}

{{editor}}'s edit

{{editor}}'s edit

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{editor}}'s edit

{{editor}}'s edit

{{title}}

{{title}}

{{editor}}'s edit

{{editor}}'s edit

Select a reply

Liquidsoap 2.3.1 (rolling release) : USB Alsa Input - Icecast Output - intermittent and juddery output (clock issues? - buffer overrun) #4236

mojikosu Dec 2, 2024

Replies: 2 comments · 14 replies

vitoyucepi Dec 3, 2024 Collaborator

vitoyucepi Dec 3, 2024 Collaborator

mojikosu Dec 3, 2024 Author

vitoyucepi Dec 4, 2024 Collaborator

mojikosu Dec 4, 2024 Author

mojikosu Dec 4, 2024 Author

mojikosu Dec 8, 2024 Author

mojikosu Dec 10, 2024 Author

toots Dec 10, 2024 Maintainer

mojikosu Dec 10, 2024 Author

toots Dec 11, 2024 Maintainer

mojikosu Dec 11, 2024 Author

mojikosu
Dec 2, 2024

Replies: 2 comments 14 replies

vitoyucepi
Dec 3, 2024
Collaborator

vitoyucepi Dec 3, 2024
Collaborator

mojikosu Dec 3, 2024
Author

vitoyucepi Dec 4, 2024
Collaborator

mojikosu Dec 4, 2024
Author

mojikosu Dec 4, 2024
Author

mojikosu
Dec 8, 2024
Author

mojikosu Dec 10, 2024
Author

toots Dec 10, 2024
Maintainer

mojikosu Dec 10, 2024
Author

toots Dec 11, 2024
Maintainer

mojikosu Dec 11, 2024
Author