[pulseaudio-discuss] audio source -> pa -> oggenc

[Maarten Bosmans]

2011/9/24 Richard Henwood <rjhenwood at yahoo.co.uk>:
> Hi Col,
>
> Just a quick message to say I've had a chance to follow up on your advice and it was helpful. Minor comments in-line, below:
>
> --- On Sun, 18/9/11, Colin Guthrie <gmane at colin.guthr.ie> wrote:
>> From: Colin Guthrie <gmane at colin.guthr.ie>
>> Subject: Re: [pulseaudio-discuss] audio source -> pa -> oggenc
>> To: pulseaudio-discuss at lists.freedesktop.org
>> Date: Sunday, 18 September, 2011, 9:38
>> 'Twas brillig, and Richard Henwood at
> <snip>
>> Add in a new sink does pretty much noting until you set
>> some (or all)
>> apps to use it, so using the Gnome sound prefs is what
>> you're expected
>> to do. Is this really problem 1?
>>
>
> Yeah, I think problem 1 is minor - and it has been resolved now I understand that I should be 'listening' to the monitor.
>
>> > PROBLEM 2: the whole task ('java StdAudio' and
>> 'oggenc') is over in a
>> > second or two. The duration of the 'sound' from
>> StdAudio is ~10
>> > seconds and what I'm aiming for is to drain audio from
>> 'java
>> > StdAudio' at real-time, not as 'fast as possible.'
>> >
>> > If you can overlook my imprecise use of terms (drain,
>> signal etc) I
>> > would really appreciate help with this :)
>>
>> Yeah the pipe sink is a bit of a hack/debug tool really,
>> and in fact
>> many distros don't ship it.
>>
>>
>> What I'd suggest is the following solution with a couple
>> variants
>> depending on how you want to use things.
>>
>>
>> The main question is: Do you want to hear the sounds as
>> well as
>> converting them to ogg?
>>
>
> No. I'm going to run my sound generating code on a server, so I don't need to hear them as well...
>
>> If you DO NOT want to hear the sounds, then you will do the
>> following:
>>
>> 1. pactl load-module module-null-sink sink_name=null
>> 2. Start "java StdAudio"
>> 3. Set the null sink to be default in the gnome GUI (or use
>> pavucontrol
>> to move only the java StdAudio stream to that device,
>> letting all other
>> sonds still use the build in h/w)
>> 4. Use the following gst pipeline: gst-launch-0.10
>> pulsesrc
>> device=null.monitor ! audioconvert  ! vorbisenc !
>> oggmux ! filesink
>> uri=file:///path/to/file.ogg
>>
>
> with: filesink location=/path/to/file.ogg
> this work perfectly - thanks!
>
> <snip>
>>
>> Also note that if you are wanting to make this available to
>> others, then
>> using the Shoutcast system is a pretty easy way to do this.
>> You can push
>> straight to shoutcast via GST:
>>
>> gst-launch-0.10 pulsesrc
>> device=alsa_output.pci_8086_27d8_alsa_playback_0.monitor !
>> audioconvert
>>  ! vorbisenc ! oggmux ! shout2send ip=SOMEIP port=8000
>> password=hackme
>> mount=stream.ogg
>>
>
> Ok, so it seems that I need to drag in shoutcast dependency :/
> It would be nice if there was a light-weight alternative to shoutcast - sighttpd might be a good place to start if I had copious free time...

I think vorbis in an rtp stream is supported by gstreamer too.

Maarten