unfortunately jack1 is not an option for me, as it's approach to pre-fault the stack does not have any checks for the stack boundaries [1] ... this consistently crashed my application tim [1] http://trac.jackaudio.org/ticket/276

if that would only be true ... a) any non-linearity introduces harmonics, some non-linearities introduce an infinite amount of harmonics, which will cause foldover distortion. the large the sampling-rate, the lower the foldover. b) delay-lines have a higher precision at higher sampling-rates c) the tuning of digital filters is more precise at higher sampling-rates due to the frequency warping in the blt iir filters may have a higher quantization noise, but that is the reason, why a good filter implementation is done in double-precision. frankly, 48k may be a good enough for distribution, but it is sub-optimal not for production ... and it is horrible for digital synthesis. fwiw, for digital synthesis (non-standard or distortion synthesis) i ended up rendering my compositions at 3mhz ... which was a good compromise between computation time and sound quality. best, tim note on a: if your signal processor introduces the Nth harmonic, you have to upsample your signal by a factor of N. or apply a pre-filter on your signal by nyquist/N. question for the reader: in order to completely prevent foldover distortion, how much do you have to upsample for a tanh waveshaper (a processor that introduces infinite harmonics)?

maybe because you did not like the aliasing artifacts and/or quantization noise? scnr

On Sun, 2014-03-16 at 18:32 +0000, Fons Adriaensen wrote: Theoretically? Plug in your Mini Moog to your RMR card and record it using 44.1 and 48. If you can't hear a difference, than visit your otolaryngologist.

Every time I'm about to jump into one of these discussions, Fons says pretty much what I was going to say. But there is one additional textbook observation I was going to make: Although this is mostly true, it's difficult to believe it matters in a practical audio application. If it does, in most of the cases that are not deliberately using a highly nonlinear transfer, adding one additional bit of depth will get you the same noise/distortion benefit as doubling the sampling rate. Once you're using 32 bit floats, you're already 160dB down... If you need more, I'd love to know what you're doing, and I'm not being sarcastic! Monty

On Sun, 2014-03-16 at 22:09 +0100, tim wrote: What happens to the noise floor, when using such a high sampling rate? The noise floor of my RME card looks disgusting at 192 KHz.

On Sun, Mar 16, 2014 at 10:37:51PM +0100, tim wrote: The F-range headroom is convenient, but not necessary. This misses the point that for 24-bit converters the noise floor will be determined by the analog electronics. In other words any noise shaping is futile. It helps only for 16-bit systems. Ciao, -- FA A world of exhaustive, reliable metadata would be an utopia. It's also a pipe-dream, founded on self-delusion, nerd hubris and hysterically inflated market opportunities. (Cory Doctorow)

On 03/16/2014 05:45 PM, tim wrote: ok, so you are trying to do weird synthesis that can produce non-bandlimited output? i can see how you might want to use high sampling rates there, but then again there will always be another processing step that causes yet higher harmonics - addressing that with high sample rates seems like a somewhat blunt approach that is bound to fail eventually. that statement is definitely not correct. granted, if you only do delays with sample granularity (which has the big advantage of not requiring any computation), there is some benefit in using higher rates. but you can produce sub-sample delays with arbitrary precision easily. for IIR feedback, i sure see the point, but then the question becomes "why do you need to expose this to the outside world?" - just upsample in your processing application and leave the rest of the jack graph running at a sane rate. i don't understand this. can you elaborate? what is "blt"? true. it's a funny and somewhat strange thought exercise for me to try and achieve the highest possible "fidelity" with brutal distortion algorithms - obviously, since i don't work with distortion, i try to keep my signal chains as linear as possible. but i can see how somebody well trained in distortion synthesis would want to eliminate aliasing artefacts, since those would conceivably interfere with systematic exploration of sounds based on prior experience, and make the sonic outcome even more erratic than it already is... but in any case, there is no point in taking the internal higher sampling rates out into the real world, so the zita resampling approach might be your best bet. incidentally, just returned from musikmesse, and i've had my share of DXD/DSD loonies... if you want to go there, there is people who want to sell you 256-times oversampled single-bit delta sigma gear, and they will happily talk megahertz with you. it would be a ton of fun to discuss with them the best way to handle a tanh waveshaper, and what new ultimate fidelity frontiers are required for the distortion synthesis crowd. just make sure you avoid the term distortion, call it "spectral enhancement processing" instead. >;-> best, jörn -- Jörn Nettingsmeier Lortzingstr. 11, 45128 Essen, Tel. +49 177 7937487 Meister für Veranstaltungstechnik (Bühne/Studio) Tonmeister VDT http://stackingdwarves.net

On 03/16/2014 09:49 PM, Fons Adriaensen wrote: ah, thanks. i had guessed that it probably wasn't "bacon/lettuce/tomato"... thanks! time to review some things... (/me grabs old hardcopy of dspguide.com) -- Jörn Nettingsmeier Lortzingstr. 11, 45128 Essen, Tel. +49 177 7937487 Meister für Veranstaltungstechnik (Bühne/Studio) Tonmeister VDT http://stackingdwarves.net

On Sunday 16 March 2014 14:25:14 Ralf Mardorf did opine: That I think is a personal call Ralf, primarily because at 48 Khz, your anti-aliasing filters had better be very very good brick walls by the time you get above 24Khz in input content just for the aliasing control. And aliasing noise, once introduced, cannot be removed by any known math function that does not have precise knowledge of the phasing (aka group delay) of the original signal. And those very good brick wall filters _will_ have a considerable group delay. IMO doing the sampling at 240K, then doing a weighted sum shift (5 stage shift) to decimate the data down to 48K, should result in dropping the alias caused noise floor by several db. Just bring lots of expensive hardware to do that. Cheers, Gene -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) Genes Web page <http://geneslinuxbox.net:6309/gene>

On Sunday 16 March 2014 15:11:20 Ralf Mardorf did opine: Pretty much true, even for golden ears, PROVIDED the anti-aliasing filtration in front of that digitizer is down at least 70 db by 1/2 the sample frequency. You would be surprised at the gear available that depends on the microphone to do 80% of the filtering, and at the atrociously digitized sound you could get when miking a set of snares or cymbals with an Altec M21b or a PZM, both of which have pretty good response yet at 25Khz. Even some of the better ribbon mikes will cover those frequencies although they would tend to be very "peaky". I hesitate to answer this, my German is far worse than your English, and I don't quite get the translation. What I am saying is that when digitizing at 48Khz, any content in the input signal that has components approaching 1/2 the digitization sampler frequency will be aliased. Said another way, a 20Khz input will be digitized as both the 20Khz fundamental, and an 8Khz beat frequency "alias" component, and that this is one of the more distracting artifacts. The only way to stop it is to filter that input to where the stuff above 16Khz has been reduced to undetectable levels. To properly decimate to 48Khz that would take a 4 stage weighted summing adder, but given that, the results might be startlingly better (at least for the golden ears) than a straight 48Khz digitization. Listening to that 192Khz recording should be a noticeable improvement compared to a 48Khz recording, just from the reduction of aliasing artifacts alone, given good enough mikes and sound sources suitable for exploring/exploiting the alias that will happen in the real world. These aliasing artifacts you don't hear during the quiet passages of the music, but will be superimposed on the higher frequency portions of the src audio signal, contaminating it as it gets louder. It will convert that cymbal to fingernails on a blackboard, ditto for the brushed snare. Cheers, Gene -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) Genes Web page <http://geneslinuxbox.net:6309/gene>

On Sunday 16 March 2014 19:56:32 Ralf Mardorf did opine: Even with tons of string squeaks from wound strings and heavily callused fingers, I doubt if that would reach high enough to be a valid test. You need a few crickets singing in the background, (typically 17 Khz) and some heavy action on a brushed snare drum to generate very much in the range of sound that would test it for sure. A "white noise" generator made out of the usual 17 stage xor gated shift register for feedback, clocked at 50Khz or more might be a good test, the white noise will seem to have a odd, often disagreeable, definitely non-harmonious pitch to it. But I haven't seen that schematic in yonks. Take a look at <http://en.wikipedia.org/wiki/Linear_feedback_shift_register> Some of the longer ones are pretty decent. the 17th register addition is one I heard once, sounded pretty white to my ears at the time. Cheers, Gene -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) Genes Web page <http://geneslinuxbox.net:6309/gene>

On Sun, 2014-03-16 at 20:26 +0000, Fons Adriaensen wrote: Many apologies :).

On 16/03/14 19:39, Gene Heskett wrote: Can anyone point out a commercially available microphone used in the audio recording domain which will actually pic frequencies above 20 kHz? Likewise can anyone point out any commercially available speaker used in the audio reproduction domain which will actually reproduce frequencies above 20 kHz? If the audio produced is made for fruition of humans it makes absolutely no sense to try and capture or reproduce anything above 20kHz, and for average individuals 15kHz would probably more than enough. And in case anyone is tempted to state that even if we don't hear them frequencies above 20kHz influence the way we hear or 'perceive' music, please also attach any _scientific_ study/paper/evidence (e.g. large-scale blind tests etc. not anecdotal evidence) to such statement. Lorenzo.

On 03/16/2014 09:50 PM, Lorenzo Sutton wrote: i once talked to a bat researcher (no joke) who had a simple mod to a røde nt5 that would allow it to work reasonably well up to 30k. earthworks make special versions of their excellent microphones which are linear up to 50khz, for those who need it (or think they need it). in fact, most if not all microphones can do this but for additional filters added by the manufacturers to increase robustness in the presence of HF electromagnetic noise. the question is how linear they are up there, and whether the pickup pattern is still useful. all but the smallest capsules can be expected to become highly directional at HF. something like a DPA 4060 or a SH MKE-1 might be good experimental capsules for ultrasonics. most tweeters are easily capable of 30khz, and there are no major engineering obstacles to 50khz on-axis. the big issue is excessive beaming, both on the recording and playback side of things. i'd tend to agree with that statement, but there are very valid reasons to do it: * not all recordings are meant for humans to hear - if you are measuring something, you might appreciate results outside of human sensation. * not all recordings are meant to be heard in its original frequency range - talk to any bat enthusiast. seriously, what those guys do makes you itch to try 192khz and a microphone that is open "from dc to daylight", as the saying goes. * sometimes, preservation of information is extremely important. for instance, there are valid reasons to digitize old analog tapes at ridiculous rates (say, 384 kHz): doing so lets you record traces of the HF bias, which might help in eliminating wow and flutter artefacts more precisely than tracking the 50 or 60hz power grid hum. or there's a colleague from italy, david monacchi, who records sound scapes in soon-to-be-destroyed natural habitats - why would you limit yourself to 10 octaves if you can get 11, before the bulldozers arrive? (i once heard him lecture on one of his works, and indeed he was using sonograms to identify certain species of animals, many of which are capable of uttering ultrasonics.) there is still ongoing debate about indirect audibility of high frequency content via transients - i'm not too convinced, but i can understand any colleague who would rather record too much today and then downsample, as opposed to finding you won't be able to fully exploit future distribution formats with your legacy material. if i'm not maxing out my equipment in terms of cpu cycles, there is no harm done in erring on the side of caution, if high sample rates don't incur higher costs as they go through the workflow. as a counter-example, a tv production i'm involved in uses 96k initially, but only because the live sound desk is a midas which cannot do less. it is immediately downsampled to 48k before going onto the broadcast network via dante, because nobody wants to put up with the extra data and doubled loadin/transfer times. actually, i just had a behringer ada8000 die on me, and i will probably replace it with some 96khz capable kit from directout. i don't expect to find anything interesting, like you said, but i'm going to try it nonetheless. i know that my tweeters will only begin to roll off at 40khz. i'm not advocating high sample rates, but hey, "because i can" has always been a strong incentive :-D the main problem is that some equipment actually gets worse at high sample rates, and putting a 192k sticker on your box is actually more important than getting 48k really right... -- Jörn Nettingsmeier Lortzingstr. 11, 45128 Essen, Tel. +49 177 7937487 Meister für Veranstaltungstechnik (Bühne/Studio) Tonmeister VDT http://stackingdwarves.net

On 16/03/14 22:34, Jörn Nettingsmeier wrote: ok ok... :-) I was obviously being a bit sarcastic in my statements.. [...] I agree with you actually, especially with all the 'preservation' scenarios... I was just being a bit provocative as to overrating gear specs.. Seriously, in all the cases you mention there is at least a thorough thought behind choice of specs and requirements for recording, and I like that. I was also thinking of all the counter-examples, in the _music_ domain, recorded on technically not-so-hi-quality gear which is just great music :-) Lorenzo.