Inside High res music: Jazz at the Pawnshop NAXOS (DSD)

[GWolfman]

Good hearing has timing resolution of about 10us, which dictates that 192 KHz samplings has to be used to capture that. On the other hand, the lower 8 bits of a 24-bit recording would likely be inaudible except in extreme scenarios: 0db noise floor and someone who is willing to pierce their ear drums with >96db sound.

But music is dynamic, so instantaneous sound over 96dB is likely common and not harmful (e.g., drum/percussion hits). Sustained notes or the wall/sound wars on the other hand are suspect.

 

[danadam]

Good hearing has timing resolution of about 10us

Yes. Some say even 5us.

which dictates that 192 KHz samplings has to be used to capture that.

No, 16/44 is perfectly capable of doing that: https://www.audiosciencereview.com/...ds/time-resolution-of-redbook-16-44-pcm.22102

 

[Wombat]

FWIW.

Recording details:

https://www.nativedsd.com/catalogue/albums/jazz-at-the-pawnshop/

 

[solderdude]

The used recorders were Nagra IV.
They used 2 of them at 15ips/38cm/s because of the small size they alternated the recorders (with some overlap).

Specs from Nagra service manual:

Freq. range (in widest range setting for low frequencies) at max speed and -20dB recording level: 30Hz to 20kHz (+/- 1dB) but of course, unlike digital it can record higher frequencies but will be attenuated. The recording amplifier has peaking at around 18kHz.
18Hz to 35kHz (- 3dB)
14Hz to 50kHz (-6dB)

bias frequency 125kHz

S/N ratio: 75dB
crosstalk at 10kHz: 50dB

 

[Ata]

No, 16/44 is perfectly capable of doing that: https://www.audiosciencereview.com/...ds/time-resolution-of-redbook-16-44-pcm.22102

Thumbs up for the reference! I stand corrected. I assume this is with a perfect reconstruction filter? Has anyone looked at filter quality vs timing resolution that can be achieved?

I have long suspected that the elaborate reconstruction/upsampling applied by software tools such as HQPlayer and hardware such as Chord M-Scaler improve soundstage by providing effectively better quality reconstruction filters, compared to a typical DACs. Now it seems that suspicion is spot on.

 

[firedog]

I don't understand the relevance or usefulness of this.
It's no secret that it's a conversion from tape. No one is being bamboozled.

If I was going to buy it, I'd buy the DXD, because it's one less generation of conversion before making the release version.

But in any case people who are buying the DSD are filtering out the ultrasonic noise on playback as part of the DSD>analog process.

And if people prefer the sound of a DSD remaster, what's wrong with that? There's a small subset of audiophiles who prefer DSD playback. That's what this site/label exists for. They also are known for doing a good job of mastering, which can be even more important than the format.

 

[Hayabusa]

@amirm
In the dsd (SACD) standard they add a 50KHz LP filter after the 1 bit dac to filter out 'HF' noise.
I assume your Adobe sw does not emulate that when showing these spectrum images?

 

[PierreV]

@amirm do you have a single example where Hi-Res undisputably brings something to the table? I haven't been able to find a single one, but that may be because of my musical preferences. I am enjoying your analyses but that may just be confirmation bias. We need a counter example

 

[voodooless]

Timing/phase information, on the other hand, can be audible as our ears use that to localise the sound. Good hearing has timing resolution of about 10us, which dictates that 192 KHz samplings has to be used to capture that.

Not this again.. I feel like on every of these videos you have to explaining again: timing resolution of a sampled signal is not discrete, therefor for all practical purposes can be seen as infinite.

192 kHz samples audio does not have more time resolution than a 44.1 kHz sampled clip within the passband of what is audible for normal humans.

 

[Frgirard]

100 pages for those pretending the 44,1 is not PERFECT.

sometime i tell me the audiophilia is a plague. Living the tech-savvy superstition.

In French we have two word for know connaître and savoir.
The audiophiles know all but know nothing

https://gearspace.com/board/music-computers/992969-88-2-96khz-vs-44-1-48khz-thread-end-them-all.html

 

[Hayabusa]

Not this again.. I feel like on every of these videos you have to explaining again: timing resolution of a sampled signal is not discrete, therefor for all practical purposes can be seen as infinite.

192 kHz samples audio does not have more time resolution than a 44.1 kHz sampled clip within the passband of what is audible for normal humans.

Timing of a sampled signal IS discrete by the quantisation to a limited number of bits.
So a 16 bit 192Khz signal has more time resolution than a 44.1KHz one.
But we are talking about pico seconds here, no where near 10 uSec.

 

[Kagemusha]

I owned the vinyl Direct to Disk. It was around $40 in the late '70s. It was a Scandanavian jazz band that was recorded live in a Stockholm night club. Was a state of the art recording in its day, but the performance left something to be desired.

Yes, done in Stockholm, the CD i listened has a great sounding to it, feels like you are in that pub, very sell recorded, why destroy it with more strange oversampling and new formats?

Some info from Wiki:
Proprius Records founder Jacob Boethius produced the album, and it has been issued at least five times under multiple labels and formats. The album is widely regarded by audiophiles as one of the best jazz recordings of the 20th Century.

 

[voodooless]

Timing of a sampled signal IS discrete by the quantisation to a limited number of bits.

Samples are discrete, the time resolution of the captured signal is not! You can encode a sine with any phase with any sample rate as long as it is sampled with at least double the sine frequency. There is nothing that says you can only capture phase in specific increments. So no, there is no time resolution advantage when sampling higher.

 

[danadam]

I have long suspected that the elaborate reconstruction/upsampling applied by software tools such as HQPlayer and hardware such as Chord M-Scaler improve soundstage by providing effectively better quality reconstruction filters, compared to a typical DACs. Now it seems that suspicion is spot on.

IMO, a "suspicion" becomes "spot on" only when it is confirmed. I don't see anything being confirmed in your post.

 

[Phorize]

Audiophile press: ‘Hi res, hurrah!’

Me: ‘Hi res, meh.’

 

[Hayabusa]

Samples are discrete, the time resolution of the captured signal is not! You can encode a sine with any phase with any sample rate as long as it is within half the sampling rate. There is nothing that says you can only capture phase in specific increments. So no, there is a time resolution advantage when sampling higher.

Only if the sample value is not discrete. If it has a fixed number of steps (2^16 for instance) all these possible steps give you the resolution to capture the phase of the signal. This is very high resultion ( approx 1/(44100*2^16) sec) but not infite.

 

[Frgirard]

CD temporal resolution an another audiophile fantasy.

https://github.com/plext/cdtr

 

[voodooless]

Only if the sample value is not discrete. If it has a fixed number of steps (2^16 for instance) all these possible steps give you the resolution to capture the phase of the signal. This is very high resultion ( approx 1/(44100*2^16) sec) but not infite.

Yes, bit depth plays its part, but the sample rate does not change that. As I said: for practical purposes, you can view it as infinite. "practical" is the magic word here.

 

[Hayabusa]

Yes, bit depth plays its part, but the sample rate does not change that. As I said: for practical purposes, you can view it as infinite. "practical" is the magic word here.

It does. for 192Khz its approx 1/(192000*2^16).

Note by the way that this for a FS signals. The number of quantisation steps that can capture the phase of a signal is less at lower ampiltudes.
Maybe we should do an experiment to show this

 

[voodooless]

It does. for 192Khz its approx 1/(192000*2^16).

No, adding more samples does not mean you gain more actual resolution.

Note by the way that this for a FS signals. The number of quantisation steps that can capture the phase of a signal is less at lower ampiltudes.

Yes, but again: the sample rate does not change that.

Maybe we should do an experiment to show this

Why? Shannon proved this long ago..