Author Topic: Sampling Rates  (Read 595 times)

Offline James Edward

  • Obsessively Audiophilic
  • ****
  • Posts: 487
Sampling Rates
« on: January 14, 2021, 02:36:47 PM »
My further adventures... Now that Im using the excellent sounding DAC within my K-07X player for streaming, curiosity ensues because it has a display...

44.1- Got that- CD

88.2- Was a CD oversampled or upsampled- explain the difference if you care to- Im all ears, pun intended. How did it get to Tidal that way?

96- Tell me more.

192- Have not seen it, nor do I expect to. Any thoughts appreciated.

Thank you
Luxman L- 590 AX MK2
Esoteric K-07X
Pro-Ject Debut Carbon Esprit SB
Audience Au24SE
PI Cat6
Spatial Audio M3 Turbo S
Hsu ULS-15 Mk2

Supra Classic 6.0

Offline HAL

  • Obsessively Audiophilic
  • ****
  • Posts: 884
  • Audio Mad Scientist
    • Hollis Audio Labs
Re: Sampling Rates
« Reply #1 on: January 14, 2021, 04:15:30 PM »
Depends on how the recording was mastered to file.  Depends on the A/D converter used for the initial recording and what the mastering engineer saved the file recording format.

Lots of music services now have HiRez capabilities to stream the higher sample rate file.  Makes no sense for them to upsample the original file as it uses more bandwidth to stream the file or download it. 

When I make recordings I use 24bit/192KHz stereo A/D converters and save as WAV files.  No reason studios cannot do it.

If it is an upsampled file, a spectrum analysis of the file with lots of free programs would show it.

Offline tmazz

  • Out Of My Speaker Cabinet
  • ******
  • Posts: 10497
  • Just basking in the glow of my tubes.....
Re: Sampling Rates
« Reply #2 on: January 14, 2021, 05:05:39 PM »
You have to separate two things, what bit rate was it originally recorded at and what bit rate are you playing it back at.

In general the higher the sampling rate used to record the music the high the quality of the reproduction. this is not unlike using a higher megapixel camera or a 45 vs 33 rpm LP. but like in both of those cases, while this looks good in theory there are always other details that can derail that. But for the moment let's assume for the sake of the example that this is true. In general, a higher bit rate will give you a smaller interval in time between the sample which will allow you to make a more accurate representation of the original signal when you decode it. The number of bits in each sample will allow you to make a finer spacing between the amplitude levels in each signal and will give you st same type of quality increase with respect top level that high sampling rates do with respect to time. Keep in mind that for any of this to work it must be done at the time of the recording.

So when you see an 88.2 indication of your DAC's display it means that the DAC is reading an 88.2 signal, but this does not tell you if the file was originally recorded at 88.2. It is not uncommon for lower bit rate files to be upsampled to a higher rate. This is done for example by taking a 44.1 CD and making a bitstream where every sample is in there twice. Each one is read twice so even though it is an 88.2 bitstream, there are only 44.1 unique samples to the resolution of this signal is no greater that the original  44.1 signal. However what this does accomplish is since the DAC is operating at 88.2 the noise artifacts generated by the digital process are pushed up higher in frequency and further away from the audio band, making them easier to filter out. Another thing that sounds very logical on paper, but there are widely varying opinions as to whether this actually leads to higher sound quality. As with anything else much of that has to do with the balance of the design other than the upsampling part. And while I just used 88.12 as an example, upsampling can be done to other higher formats as well.

So in general, high sampling rates and longer word lengths can lead to high quality. But keep in mind that they also lead to very large jumps in file size. And like anything else in this hobby you quickly get to the point where the law of diminishing returns kicks in. All things being equal, an 88 or 96 file will sound better that a CD. 1 992 file can sound better than a 88, as will every other step up the chain, but by how much, and at what point does the extra size and cost of the file start to exceed the value of the increased SQ? Like in all of our other decisions that is the $64,000 question. There are plenty of people who think we are all crazy and an mp3 file file played through a set of $5 earbuds is all you should ever need . And we all know how nuerotic some of us can get about SQ in all aspects of this hobby. So the choice is really up to you.

Personally I have found to my taste, given the equipment currently in my system, there is not much to be had by going above 192. I do have a lot of DSD files, but that is mostly because I already had a lot of SACD that I was able to rip the high res layer from to play via Roon. So for me that decision was made more based on availability that SQ.

If you are streaming remember that while Qobuz will stream files to you in their native high res formats, up to 192, if Tidal is your service of choice they cannot stream anything higher than 44.1 and anything you see as a higher bitrate on your DAC is the result of some kind of MQA decoding in your system. (The Bluesound Nodes will do the fist layer of MQA software unpacking and then output a digital signal of up to 96/24 to your DAC. If you have hardware MQA in your DAC it will support MQA unpacking up to higher bitrates.)

Blue Coast Records at one point has a free download of a song by Keith Greeninger called Close to the Soul that was given to you in every digital format from CD all the way up to DSD 512 so you could hear the differences with your own ears.

Remember, it's all about the music........

Nola Boxers
Sunfire True SW Super Jr (2)
McIntosh MC 275
ARC SP-9
VPI HW-19 Mk IV/SDS/SME IV/Benz Micro Glider
Pro-Ject Pre Box S2 DAC/Rasp Pi Roon Endpoint/Denon 2910
DigiBuss/TWL PC&USB/MIT Cable

Offline rollo

  • Industry Participant
  • *
  • Posts: 6407
  • Rollo Audio - Home demo the only way to know
Re: Sampling Rates
« Reply #3 on: January 15, 2021, 08:01:56 AM »
  Sound quality of 44.1 recordings upsampled varies. Most IMHO do not sound better. If the recording was recorded in Hi-Rez then played back in Hi-Rez an improvement will be heard. On QoBuz you can stream Hi-Rez however do not know if recorded in Hi-Rez or upsampled. Bluecoast offers 24/192 on QoBuz that was recorded at 24/192. Go to Qobuz playlists then to Hi Fi partners there you will find lots of Hi Rez especially Bluecoast.


charles
contact me  at rollo14@verizon.net or visit us on Facebook
Lamm Industries - Aqua Acoustic, Formula & La Scala DAC- INNUOS  - Rethm - Kuzma - QLN - Audio Hungary Qualiton - Fritz speakers -Gigawatt -Vinnie Rossi,TWL, Mad Scientist, Swiss Cables

Offline James Edward

  • Obsessively Audiophilic
  • ****
  • Posts: 487
Re: Sampling Rates
« Reply #4 on: January 15, 2021, 02:26:44 PM »
Long week at work- Im not getting any younger. I appreciate the replies and will digest and reply tomorrow morning.
Thank you.
Jim
Luxman L- 590 AX MK2
Esoteric K-07X
Pro-Ject Debut Carbon Esprit SB
Audience Au24SE
PI Cat6
Spatial Audio M3 Turbo S
Hsu ULS-15 Mk2

Supra Classic 6.0

Offline Brap

  • Obsessively Audiophilic
  • ****
  • Posts: 273
Re: Sampling Rates
« Reply #5 on: January 17, 2021, 10:53:47 AM »
Jim,  just checking to see if you got my PM.
Fritz Carrera 7 Be
VTA ST-70
Parasound Halo P5
Oppo 105
Rega P3
Ortofon Quintet Bronze
Cambridge CXN streamer
Teac 3340S
Anticables 3.1
Masterbuilt reference/Kimber Cable IC

Offline tmazz

  • Out Of My Speaker Cabinet
  • ******
  • Posts: 10497
  • Just basking in the glow of my tubes.....
Re: Sampling Rates
« Reply #6 on: January 18, 2021, 02:24:41 PM »
Here is a link to an article discussing probably more than you ever wanted to know about digital bit rates and formats:

https://www.mojo-audio.com/blog/dsd-vs-pcm-myth-vs-truth/
Remember, it's all about the music........

Nola Boxers
Sunfire True SW Super Jr (2)
McIntosh MC 275
ARC SP-9
VPI HW-19 Mk IV/SDS/SME IV/Benz Micro Glider
Pro-Ject Pre Box S2 DAC/Rasp Pi Roon Endpoint/Denon 2910
DigiBuss/TWL PC&USB/MIT Cable

Offline James Edward

  • Obsessively Audiophilic
  • ****
  • Posts: 487
Re: Sampling Rates
« Reply #7 on: January 18, 2021, 05:04:58 PM »
Jim,  just checking to see if you got my PM.
Just replied...
Luxman L- 590 AX MK2
Esoteric K-07X
Pro-Ject Debut Carbon Esprit SB
Audience Au24SE
PI Cat6
Spatial Audio M3 Turbo S
Hsu ULS-15 Mk2

Supra Classic 6.0

Offline Nick B

  • Global Moderator
  • *****
  • Posts: 2582
Re: Sampling Rates
« Reply #8 on: January 18, 2021, 06:51:45 PM »
Here is a link to an article discussing probably more than you ever wanted to know about digital bit rates and formats:

https://www.mojo-audio.com/blog/dsd-vs-pcm-myth-vs-truth/

Thanks for posting. A very interesting and informative article. I bought a power supply from Mojo a few years ago for my Auralic Mini and it was well worth the money.
ICEpower 1200as2 amp
Audio Hungary APR 204 preamp
Fritz Carrera 7 BE speakers
Border Patrol SEi dac
Auralic Aries Mini & Mojo Audio lps
Audio Envy cables
Roon, Tidal, Qobuz
PI Audio UberBUSS

Offline steve

  • Audiologist
  • *
  • Posts: 843
    • Reference components
Re: Sampling Rates
« Reply #9 on: January 18, 2021, 11:29:26 PM »
Of course I have to stick my nose in, for a few points.

Below is the reproduction of a 3us (microsecond) analog duration pulse. Only DSD properly reproduced the pulse.
The others lacked not only amplitude accuracy, but severe ringing.

One may say, but only 3us width pulse. I know of at least two studies demonstrating 5us is perceivable/brain
perceived, Jneutron (FermiLab, Cern, Brookhaven National Lab) noted research of 2us rise time being perceived.

The other point has to do with "blind testing or double blind listening testing". The problem/marketing hype lies in
the false belief that "sight" is the only confound variable that needs to be dealth with. That is completely false.
Just one other, simple confound variable is "cochlea fatigue". The studies/claims are just not rigorous by any means.

Question. How does one arrive at 95% confidence when statistically half of test subjects are in bass increasing
mode areas and half are in bass decreasing mode areas at the testing venue?

The use of typical double blind testing has become too common, without understanding the basic principles involved.

cheers

steve
« Last Edit: January 18, 2021, 11:44:49 PM by steve »
Steve Sammet (retired, owner, SAS Audio Labs)
"V" Very Low Capacitance ICs, 40pf 1 meter
SAS Audio Labs Test Phono Stage
SAS Audio 11A Tube Reference Preamplifier
SAS Audio Labs 25 W Triode Reference Monoblocks
2 way test Spkrs, 28 - 20khz  -3db (28hz)

Offline HAL

  • Obsessively Audiophilic
  • ****
  • Posts: 884
  • Audio Mad Scientist
    • Hollis Audio Labs
Re: Sampling Rates
« Reply #10 on: January 19, 2021, 05:23:19 AM »
Most modern PCM DAC's now have other than linear phase filters that do not have pre-ringing as shown in the picture.  They also have much less time ring that does not smear out the pulse. It is odd the pulse shown has signal before the rising edge from the analog generator.  Also if that was an analog pulse, it would have gone through both ADC and DAC.  If you want to see just the DAC. a single sample of 1 file of 0's at the sample rate would have been a better test.  PCM is not usually that inaccurate.  I worked with DSP based test equipment with DAC's and the signal output was much more accurate than that. 

It would be good to know where that waveform picture is from for reference.

Also putting ab analog 3uS pulse with a 333.333KHz bandwidth through an anti-aliasing and reconstruction filter for 22KHz rolloff will do that.  Don't think I hear that high anyway.



« Last Edit: January 19, 2021, 05:38:25 AM by HAL »

Offline steve

  • Audiologist
  • *
  • Posts: 843
    • Reference components
Re: Sampling Rates
« Reply #11 on: January 19, 2021, 11:28:51 AM »
Most modern PCM DAC's now have other than linear phase filters that do not have pre-ringing as shown in the picture.  They also have much less time ring that does not smear out the pulse. It is odd the pulse shown has signal before the rising edge from the analog generator.  Also if that was an analog pulse, it would have gone through both ADC and DAC.  If you want to see just the DAC. a single sample of 1 file of 0's at the sample rate would have been a better test.  PCM is not usually that inaccurate.  I worked with DSP based test equipment with DAC's and the signal output was much more accurate than that. 

It would be good to know where that waveform picture is from for reference.

Also putting ab analog 3uS pulse with a 333.333KHz bandwidth through an anti-aliasing and reconstruction filter for 22KHz rolloff will do that.  Don't think I hear that high anyway.

Ringing is an inherent problem with a "brick wall" filter. One of the basic reasons for upsampling is to use a lower order filter/more gradual filter in order to minimize such ringing.

2 microseconds, as Jneutron has investigated in my last post, is even higher frequency than the other two examples I mentioned previously. Higher than 20khz is perceived when accompanied by an under 20khz signal as it alters the rise time (attack time to audiophiles), the slope of the musical signal. Every time we add a stage to a musical system, rise time/attack time degradation occurs.

If the ringing did not occur as shown, I would have to question the validity of such a response.

Where I obtained the response data, was over a decade ago, and I cannot remember. However, it was not debated by other experts at Stereophile Forum, which included John Atkinson and other engineers.

cheers

steve
« Last Edit: January 19, 2021, 12:21:52 PM by steve »
Steve Sammet (retired, owner, SAS Audio Labs)
"V" Very Low Capacitance ICs, 40pf 1 meter
SAS Audio Labs Test Phono Stage
SAS Audio 11A Tube Reference Preamplifier
SAS Audio Labs 25 W Triode Reference Monoblocks
2 way test Spkrs, 28 - 20khz  -3db (28hz)

Offline HAL

  • Obsessively Audiophilic
  • ****
  • Posts: 884
  • Audio Mad Scientist
    • Hollis Audio Labs
Re: Sampling Rates
« Reply #12 on: January 19, 2021, 04:10:44 PM »
Seeing a difference in a 333KHz to 500KHz waveform and hearing a difference is very different.  Pre-ringing in a linear phase filter is a non-causal process.  All the waveforms except the DSD one shows pre-ringing.  For the last 10+ years there have been minimum phase reconstruction filters in DAC's that do not have pre-ringing and sound much better than what is shown.  Once those have been used and sample rates of 192KHz, I hear no difference to SACD(DSD) file replay Natively. 

So if there was a listening test, would anyone hear the difference after getting rid of linear phase filters and using a sample rate of 192KHz and 24bit quantization?  That is the more important question to me.  With DSD being a proprietary process, I find it much less expensive and excellent sound quality.

YMMV.

Cheers

Offline James Edward

  • Obsessively Audiophilic
  • ****
  • Posts: 487
Re: Sampling Rates
« Reply #13 on: January 19, 2021, 04:24:23 PM »
Can someone put forth some things to investigate regarding why my CDs generally sound better through the same DAC Im using for streaming. It is literally in the same box. I am using the DAC inputs within my player- K07-X. Is it possible that the digital cable is at fault? It is the specified 75 Ohm interface, though not a boutique brand.
What else is at play?
My ears tell me that ordinary cds sound better than 88.2 or 96 kHz streaming. In theory, shouldnt higher bit rates equate to better sound?
My list of culprits:
The digital cable
The streaming service itself- Tidal
Any thoughts appreciated.
Jim
Luxman L- 590 AX MK2
Esoteric K-07X
Pro-Ject Debut Carbon Esprit SB
Audience Au24SE
PI Cat6
Spatial Audio M3 Turbo S
Hsu ULS-15 Mk2

Supra Classic 6.0

Offline HAL

  • Obsessively Audiophilic
  • ****
  • Posts: 884
  • Audio Mad Scientist
    • Hollis Audio Labs
Re: Sampling Rates
« Reply #14 on: January 19, 2021, 06:33:56 PM »
James,
If you are using an S/PDIF coaxial interface between the streamer and DAC, then that is a possible cause.

Internally most CD players use a direct data and clock connection called I2S Bus after decoding the CD players player.  It can have a very low noise clock system for low jitter.  S/PDIF dending on how it is implemented, can have widely varying jitter on the interface.

As a start test, you might want to try just try an RG59 75 Ohm RF cable between the streamer and DAC.  If it sounds different, then that can point to the coaxial interface as the source of the sound change between the CD and streamer sound.