192khz 48khz recording

[ruffrecords]

Does this not simply demonstrate that there is absolutely no more information in a 24/192 data stream than there is in a 24/48 stream if the sampled signal is below 48K/2 .


Cheers

Ian

 

[mattiasNYC]

exactly

 

[JohnRoberts]

I think people should actually loosen the boundaries of audible sound a bit. When doing a simple tone sweep I can't hear above 17K  as a guy in his 40's. My kid says he might be able to "feel" the tone a bit above 18K if its loud enough.  But that's still 2000hz away from 20K. 

The limits of audibility are well studied but is relatively easy to just deliver 20-20kHz and not worry about it.

Does it really matter if there is inter-modulation distortion around 19 or 20K anyhow? Maybe thats the "air" people hear sometimes. Maybe crappy converters with bad filters will come back like vinyl did!

IMD is not "air" more like LF grunge...

Back in the '70s I modified my old heathkit SPMTE IMD tester to use 19kHz and 20kHz, instead of the stock 7kHz and 60Hz that was way too easy even back in the 70's. The IMD distortion product from 19k&20k was a very audible 1kHz.

This two tone IMD was a revelation when designing high performance RIAA preamps (back when vinyl was still a thing), instead of the standard THD+N in common use. Harmoninc distortion was rolled off by the RIAA LPF pole at 75uSec, and the falling GBW of 70's era op amps, but the 1kHz IMD product received 20dB of boost relative to the 19/20kHz stimulus due to that same RIAA eq,  so preamps that looked clean with THD testing , were much less clean when looking at their HF  IMD. Note SMPTE IMD looked even cleaner than 20khz THD because of the modest 7kHz stress, but 19&20kHz makes them work, and the distortion is all in band so no magical thinking or funny business. 

This is fairly old news now, but new to me in the '70s

P.S. Anyone going down the sample rate rabbit hole should read Dan Lavery's papers.

http://www.lavryengineering.com/pdfs/lavry-white-paper-the_optimal_sample_rate_for_quality_audio.pdf

Dan Lavry's work wrt sample rate and a/d conversion is very well respected , but I doubt he discounts HF linearity (IMD). 

JR

 

[bluebird]

Oops corrected the spelling.

Thanks John, I was just asking as I don't know much about IMD or its effects. Looks like I should study up on it a bit.  I just got an AP system one and believe there are tests you can set up for testing IMD.

Do you know if low pass filters in digital conversion cause IMD? Or just resonant peaks around the cutoff frequency?

 

[JohnRoberts]

Oops corrected the spelling.

Thanks John, I was just asking as I don't know much about IMD or its effects. Looks like I should study up on it a bit.  I just got an AP system one and believe there are tests you can set up for testing IMD.

Do you know if low pass filters in digital conversion cause IMD? Or just resonant peaks around the cutoff frequency?

IMD is caused by non-linearity... When a pure tone is distorted by non-linearity it creates (higher) harmonic overtones. When multiple tones are distorted by non-linearities the distortion products are (higher and lower) sum and difference frequencies.

I think the AP supports two tone IMD and even more exotic multiple tone IMD tests.

I wouldn't expect sample rate to affect linearity.

JR

 

[john12ax7]

I've done these tests in the past using the same source material and could hear and measure a difference.  Converters have a sweet spot where they sound and perform best,  but it's not necessarily the same for each converter.

88.2/96 work better for me than 44.1/48. Also,  in my case 192 was worse than 96.

 

[Script]

Yeah, if target is 44.1, I record/process at 88.2. Then final downsampling is an easy mathematical operation. Maybe it's magical thinking, but I tend to believe this to be better than converting 96 to 44 or 88 to 48.

When heavy ITB processing is required (such as full-fledged audio restoration), I regularly check the frequency range above 20/22KHz in spectral analysis between steps and sometimes even clean it up by hand should there be any weird stuff going on (e.g., clipping from original recording extrapolated into higher freq range). I don't blindly trust plugins to do the necessary filtering well.

Also, just like with any hardware gear, I always A/B any new plugin at zero settings before use. Does the plugin have a sound of its own? If so, do I want that? And is that 'mojo' or just 'peyote'?

 

[kambo]

Also,  in my case 192 was worse than 96.

i made the same test in the past too, and results were close to ur results.
timestretch algorithms are getting better for sure.
currently, 32bit /192 khz is the worst results on my end!

 

[JohnRoberts]

timestretch algorithms are getting better for sure.

The gold standard for pitch shifting back in the day involved what was called a "rotating head" tape machine. Four(?) heads were wired in parallel and spaced around a revolving drum that was in contact  with the moving tape. By varying the rotation velocity of the head assembly you could shift the pitch of the recorded material up/down. (This is similar to how VCRs work but for a different reason, VCRs need to trick the tape/head into high velocity relative to the actual tape speed to record and playback the HF video content. )

These parallel rotating heads rather elegantly dealt with the problems caused when splicing audio  samples together, the adjacent magnetic heads would pick up the longer wavelength signals before the shorter HF content, resulting in some bass overlap and less/no HF glitches as subsequent samples melded together.  Some modern pitch shifters try to mimic the splicing behavior of the old rotating head machines.

I heard rumors about some massive research projects at Bell Labs involving FFT conversions back and forth but never saw/heard any of their results.

JR

 

[Matt Nolan]

Daisy, Daisy. Give me your answer do...

 

[kambo]

The gold standard for pitch shifting back in the day involved what was called a "rotating head" tape machine.
JR

i often chop short clip of audio, then  copy/paste/reverse (optional pitch change/pitch curve) number of times...
takes more time, but works better in some situations!