hello
in a 192 kHz session, I would need a 48 kHz output buss to route signal to a device only accepting 48 kHz ( real time external FX )
as Im looking for different ways to do this, many thanks for new ideas
Sorry i don’t know if there is a way to do that other than doing it through a D/A A/D external conversion, in any case i don’t see why i personally would yo need to have a session at 192 kHz unless it is to experiment or show off to a audiophile client, there are dozens of other things in the audio chain that can flaw a recording that would actually make a noticeable difference if taken care of rather than the lower sample rate, actually 48 kHz is proven to be way more than enough, even 44.1 but considering the Video standard is 48 kHz then working at 48 kHz is a simple and smart way to save time in several scenarios.
Sorry again for going out of your thread to the sample rate topic, couln’t avoid it.
JL: run alsa_out or zita_j2a which can resample. Note that neither were written with the intention of such drastic resampling - their resampling is intended primarily to keep time sync - so they may blow up at such an attempt.
Frankly, mixing SR’s within a digital audio setup is generally an extremely bad idea.
If I understand correctly, then I suspect this is a problem you would need to solve in hardware (JACK connects to one audio I / O interface, and runs at the sample rate set by that interface, so you wouldn’t be able to split outputs to different sample rate I / O easily via JACK, even if it were possible to do some kind of SRC within ardour). You could route a single buss to a different physical I / O and then perhaps set that I / O to do SRC to 48K in the interface (depending upon the interface). Otherwise, why do you need the session at 192kHz? (is this for some format specific reason, or just due to some notion of higher quality? If the latter then it is almost certainly better to run / convert the session to 48K as the actual audible difference is in most cases negligible).
to all, thank you so much for your help
Ernesto, linuxdsp, Ive tried all sampling rates; the goal is to get the lowest latency for real time feeling when playing the pianoteq keys 192kHz 512 or 1024, ( close to round 3ms ) …and to avoid xruns …
Paul , Ill try what you write
to be short, even if its sounds crazy: I would like to use the KDFX in my K2500 for the reverbs, and the KDFX digitoul I°O only allows 48kHz
in the past I always used 48kHz, to stay compatible… all FXs in KDFX are fine, and as long I could stay in 48kHz, I used them, ( bgates, multiband comps, reverbs…) with complex I°O routing ( sorry Paul ) ;)… as my records are mostly acoustic instruments multitracks ( pianoteq for pre-rec ), 192 kHz is good (!)
thanks again
forgot: to linuxdsp:
one of the solution is, as you wrote ( Ive done that) 2 channels analog out to in K2500 and digital back to an Ardour track
@JL: ok, I can appreciate the latency issue, however, while you may be able to use high sample rates to get low latency with larger buffer settings, I don’t understand quite how that helps, because xruns are caused by buffer over / underruns, and by setting the sample rate higher, you just process through the available buffer quicker (512 at 96K would be like 256 at 48K for example) and possibly with slightly greater CPU overhead depending upon the system / interface.
As regards quality - again, there’s little to be gained at higher rates, because, ignoring bit-depth or sample resolution issues, if you sample a band-limited signal at greater than twice its highest frequency, you can completely reproduce it, so even at 44.1 you can completely reproduce any signal within the audible range. Higher sample rates just mean you burn through disk space faster when recording and have less DSP available for plugins etc during mixing / playback. (and if you are having to process any part of your session through some part of the signal chain which is at a lower rate, e.g. external FX, at precisely that point, haven’t you negated the - small - advantage of having the rest of the session at a higher rate?)
this isn’t about “complex routing”. it is about (a) mixing different SR’s in the same digital audio graph and (b) using an unnecessary and unjustifiably high SR to begin with. 192kHz has no verifiable benefits. You will find no documented, tested evidence anywhere that says otherwise. It is marketing snake oil, pure and simple. As LinuxDSP wrote, you can get the same latency with more sensible SR’s …
@linuxdsp I had these considerations you write about; but Ive tested all possible sample rates from 44,1 to 192 and latency from 64 to 1024…
keeping 192 512 2,7 ms ( 1024 giving 5,3 ms ) i noticed that one Midi track cause xruns, and the 192-512 is almost xruns free ( Ive posted bugtracker ireports )
I dont believe the hardware is the problem , and as Paul asked, the forum is not the right place for more details ( OS X, GUI ), please have a look on bug tracker, youll see the hardware details
Ok with your quality regards… M. Nyquist is not far away :)))) … we can hear 20kHz… 40kHz are sufficient…
Im not sure, but this is another discussion and would bring us on very distant ways… :)))
but sure, I understand what you mean and after all these questions… our music will be listened in mp3 … 
…while were looking for 0,003dB in the fifth harmonic or transient…
thanks Paul, some elements in my answer to linuxdsp
you wrote:and simple. As LinuxDSP wrote, you can get the same latency with more sensible SR’s …
sorry: Ive tried…
…to stay in 48 kHz,
in qjackctl ( under OS X.8.5)
with frames/period value : 2048
Sample Rate : 48000, its possible to get xruns free
5 stereo tracks
48 kHz 32 floating—latency 128 2,7 ms
6 stereo tracks ,
48 kHz 32 floating—latency 256 5,3 ms
sure not limited to 6 stereo tracks… was just for test…