Re: [LAU] [Bulk] Re: How to run jack at 8kHz sr? - Linux-audio-user

18 Sep 2015

On 17/09/2015 23:13, Alexandre DENIS wrote:
...
  On Thu, 17 Sep 2015 18:57:08 +0200
 Alessio Degani &lt;alessio.degani(a)ymail.com&gt; wrote:

> On 17/09/2015 18:01, Alexandre DENIS wrote:
>> On Thu, 17 Sep 2015 17:08:22 +0200 (CEST)
>> Tim Goetze &lt;tim(a)quitte.de&gt; wrote:
>>
>>> [Alessio Degani]
>>>> What can I do to test my algorithm at 8 kHz in realtime using
>>>> jack?
>>> Or simpler : use the 'dummy' backend for jackd at 8kHz, and
>>> connect your soundcard using zita-a2j/j2a. -a.
> Good hint!
> Humm... never thougth to use the "dummy" driver. Hi,
...
  Hi,

 It's a classical trick to decouple jackd from device, e.g. to make a
 jack session survive an USB device disconnect.

  It seems to work, but I see aliasing :) 
Isn't it precisely to investigate these issues that you wanted to try
 8kHz? Actually no... I want to test if the algorithm works equally well (at 
least in the reduced bandwidth given by the 8 kHz sampling) with respect 
to the 44.1 kHz sample rate case!
That's becouse the target hardware has a low sample rate, but with a 
proper anti-aliasing filter.
For now, I'm designing the algorithm on a PC with jack mainly for 2 reasons:
1- Debugging is far more simple
2- I don't have the target hardware yet :)

...

  Maybe I've to put a ladspa low-pass near 4
kHz just after my signal
 source?  Of course you cannot output anything higher than the Nyquist frequency.
 If your source is higher frequency than output and you need
 downsampling, a brickwall low pass filter is necessary. Sure... I was thinking that
jack 'dummy' driver takes care of 
resampling, but clearly that's not the case (as I mentioned before, I've 
never used the dummy driver before :) )
...

 BTW why did you reply privately and not to the list? Sorry... my fault! My email
client WAS configured to 'reply to list' by 
default, but for some reason something went wrong!

Cheers

-- 
a.