Re: [Jack-Devel] Problems with audioadapter used with capture only interface

On Sun, Dec 29, 2013 at 10:45 PM, Jeremy Jongepier
<[hidden]> wrote:
> On 12/29/2013 10:38 PM, Wojciech Zabolotny wrote:
>> 29-12-2013 19:32, "Jeremy Jongepier" <[hidden]> napisa³(a):
>>>
>>> On 12/29/2013 01:27 AM, Wojciech Zabolotny wrote:
>>>> Hi,
>>>>
>>>> I wanted to use jack2 with an embedded system equipped with capture
>>>> only interface.
>>>> I can start the server with the net backend:
>>>
>>> Hello Wojciech,
>>>
>>> Why the net backend? It seems as if you want to use the capture only
>>> interface on a standalone embedded device.
>>>
>>> Best,
>>>
>>> Jeremy
>>>
>> Hi Jeremy,
>>
>> the whole design is a WiFi based wireless guitar system.
>> Therefore the audio samples must be transmitted to the jack running on a PC
>> working as a virtual amplifier and the Digital Audio Workstation.
>> In fact I'm still considering getting rid of jack at the embedded system
>> side. From the latency point of view the optimal solution would be to
>> transmit the data directly from the soundcard via net to something like
>> modified alsa_in on the PC. I have even considered making of virtual ALSA
>> card with network input.
>> Regards,
>> Wojtek
>
> Hi Wojtek,
>
> Thanks for the explanation! Sounds like a cool idea. Why not run the
> virtual amp directly on the embedded device? Not enough processing
> power? And of course you could try using ALSA directly but from personal
> experience I can say that using JACK might just be the easiest option.
>
> Best,
>
> Jeremy
>

Hi Jeremy,

The whole idea was started as an upgrade of my old system:
 http://www.ise.pw.edu.pl/~wzab/wireless_guitar_system/index.html
The whole device must be small, and either attached to the guitar itself,
or to the strap, so that the guitarist can conveniently select tone
and make some
settings adjustments.
The device should also consume as small power as possible, so that it
can work long from small and lightweight batteries.
That's why no virtual amp in the device itself.
Current version of hardware is available at:
https://github.com/wzab/raspberry-pi-guitar-system-v1/tree/master/kicad-project-v1

As the latency problem is essential, I'm simply afraid that in the
configuration:
Codec-ALSA-Jack-Net-Jack the latency will be introduced twice - in
both jack servers.
In the ALSA based solution, the samples could be collected and
transferred immediately
as they are available.
At the receiving side, I considered either something based on snd-aloop
with user-space application receiving data from the network
or something like minivosc:
https://github.com/stadaki/alsa-minivosc-src
http://www.alsa-project.org/main/index.php/Minivosc
The latter would be especially efficient with private kernel-space implemented
lightweight reliable network protocol (e.g. something like what I've created for
FPGA based DAQ systems - http://opencores.org/project,fade_ether_protocol )
The data from virtual ALSA card will be received by standard alsa_in
or audioadapter
(which of course will introduce some additional latency due to synchronization).
-- 
Regards,
Wojtek

1388355387.3447_0.ltw:2, <CAMduVEdJ8Khr7JB=0diyfHsL5_OBO+HV865NardmVs_7ARB7wQ at mail dot gmail dot com>