AudioNervosa

Systemic Development => Digital Audio Devices => Streaming Players => Topic started by: rollo on July 31, 2017, 01:17:12 PM

Title: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: rollo on July 31, 2017, 01:17:12 PM
  I say a big flat no. Leave it be. To date every resampled/upsampled piece of music I have heard sounds Hi/Fi to me. Now music recorded in DSD or Hi/Rez then upsampled is better but still no cigar IMHO.
   Chew on that awhile and respond accordingly.


charles
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: tmazz on August 14, 2017, 08:33:47 AM
Are you talking about an upsampled file, or hardware upsampling of a redbook file ad part of the D/A process?

I can see no good reason for upsampling a file in the mastering process since no matter what speed you distribute the file at, it can never have more resolution that the original source.

Now if you are talking about upsampling within a DAC that is another question. If there is a switch on the DAC that enables or disables the upsampling I would simply try it both ways and stick with the one that sounds better to you. But unfortunately there are not a lot of DACs that I know of that have this feature. So if the upsampling is not defeatable it becomes just design choice made by the designer and just like all of the other design choices made (chips, op-amps caps etc.) simply becomes part of the overall sound of the DAC. Since there are so many other things that could effect the final sound it is very tough to isolate the impact of upsampling alone. By limiting yourself to only oversampling or non oversamplg DACs you run the risk of missing a DAC that even though it is outside of  your thoughts about preferred sampling, had better designs in other areas such that it sounded better that DACs using your preferred sampling philosophy.

Pre-judging equipment based on technologies can lead to a lot of missed opportunities. Hey, based on all conventional wisdom switching amps should be limited to industrial applications where sound quality is much less important than size and price.But I dare anyone who has listened to the Arions or one of Tommy's Cherry amps to say that Class D has no place in a high end system.
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: mikeeastman on August 14, 2017, 09:40:34 AM
I have to completely disagree, since I upgraded my front end and up sample everything to DSD156 there are rebook files that I could not listen to before ( I got spoiled listening to hi res files ) now most of them have never sounded so good, Still very badly recorded one still don’t sound that good, but better than when not up sampled.
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: richidoo on August 14, 2017, 09:58:36 AM
Old style hardware upsampling with built in IC never did much for me. It might have sounded "smoother" but also stuffy and dead or boring. The life and edge were reduced.

But the recent advent of excellent low jitter hirez DACs and cheap and powerful music computers spawned the emergence of great software for interpolative upsampling, or "resampling."  The steppy waveform of 44.1kHz is redrawn into a smooth analog like waveform, using advanced algorithms to fill in the missing music between the input digital samples.

SoX (http://sox.sourceforge.net/SoX/Resampling) is one resampling algorithm that I've used, it comes included with Moode Audio software for Raspberry Pi. Sounds excellent when resampling from 16/44.1kHz to 32/384kHz. Slightly soft, but most people like that. The algorithm settings are not easily adjustable in Moode.

Even better sounding is Roon  (https://roonlabs.com/)upsampling DSP, running on Windows PC, outputting to SoTM network audio adapter thingy. This sounds really good, smooth and musical, but clearer than Sox with Moode settings. Upsampling settings in Rune aren't much adjustable either.

Another I want to try is HQPlayer (http://www.signalyst.com/consumer.html) software on the PC, which also outputs to the NAA. Supposed to be even better than Roon, much more adjustable algorithm settings. Tweakers delight.

As Mike said, these new high end resampling algorithms can interpolate 44.1kHz input to any high rez sampling speed, even DSD512, which is like 5GHz? Resample to whatever is the highest native resolution of your DAC, so you can eliminating the lesser SQ onboard upsampler in the DAC IC or separate upsampling IC before the DAC in older units.
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: tmazz on August 14, 2017, 12:10:36 PM
Sounds like there are some recent developments that I have not kept up to date with.

It may be time to do some investigation.........
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: shadowlight on August 14, 2017, 02:11:56 PM

Another I want to try is HQPlayer (http://www.signalyst.com/consumer.html) software on the PC, which also outputs to the NAA. Supposed to be even better than Roon, much more adjustable algorithm settings. Tweakers delight.


I use HQPlayer Embedded on Linux with Windows based NAA and JRiver as library manager to up sample everything to 512 DSD.  The only thing that I have to be careful about is making sure that I the appropriate filter for my HQPlayer Embedded PC, otherwise I end up with stuttering.  For now I am sticking with -2s filters.  Would like to try the xtr filters but not enough horses left in the PC. 
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: _Scotty_ on August 14, 2017, 06:32:37 PM
With alternatives available in DAC hardware and software it makes sense to try multiple software combinations of upsampling as well as software based filtering options such as found in HQPlayer to get the best sound out your particular mix of gear.
 It is possible that I might get a net overall gain in sound quality if I had a computer with the horse power to run HQPlayer and upsample all PCM to DSD 128 before sending on to the Auralic Vega.
 As it is, software upsampling via jRiver audibly degrades playback when compared leaving the base sample rate of the data unchanged. Allegedly DSD is handled completely differently from PCM which is upsampled by an ARM 9–based Sanctuary processor running at 500Mips and is used in the Vega to upsample PCM input data to approximately 1.5MHz and 32-bit depth. The ESS Sabre32 9018 D/A converter chip will accept data of sample rates up to 1.536MHz and 32 bit depth. DoP v1.1 protocol and DXD 24-bit/352.8kHz PCM format is used somewhere in there for DSD playback. With as much signal processing going on internally in the Vega as there is, I guess I am not very surprised that more signal processing done externally doesn't appear to offer any sonic advantages.
 I will have to try encoding the PCM data to DSD native format via jRiver and see if that sounds better or worse than leaving it unchanged.
Scotty
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: tmazz on August 14, 2017, 07:01:32 PM

Another I want to try is HQPlayer (http://www.signalyst.com/consumer.html) software on the PC, which also outputs to the NAA. Supposed to be even better than Roon, much more adjustable algorithm settings. Tweakers delight.


I use HQPlayer Embedded on Linux with Windows based NAA and JRiver as library manager to up sample everything to 512 DSD.  The only thing that I have to be careful about is making sure that I the appropriate filter for my HQPlayer Embedded PC, otherwise I end up with stuttering.  For now I am sticking with -2s filters.  Would like to try the xtr filters but not enough horses left in the PC.

Oh boy do I have a lot of catching up to do. Damn, I don't think I would understood any less of Deepak's reply if it was written in ancient Babylonian.  [-(
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: Nick B on August 14, 2017, 09:43:29 PM
I've enjoyed the upsampling capability of Roon. I have it set at 24/192 and all my CDs are 16/44. Roon will upsample to DSD iirc, but my dac can't handle that. I would love to hear DSD in my home
Nick
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: rollo on January 25, 2018, 09:24:04 AM
  OK guys you have given me reason to try , listen and make my decision. My DAC is capable of handling incoming up to double DSD. When I play upsampled on the Innuous sometimes it sounds better. Other times sounds Hi Fi. So far 24/192 from server is very good. DSD when originally recorded as such has potential.
Without sounding like an Arse 44.1 done right is hard to beat. However the "Old Horse", me will give it another try.


charles
 
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: Nick B on January 25, 2018, 09:50:47 AM
Would be interesting to hear from Dave and Pete as they are NOS guys that use a Mac mini. What is their opinion of DSD 512 and do they think upsampling has any redeeming qualities.
Nick
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: _Scotty_ on January 25, 2018, 09:58:21 AM
Charles, if you are going to upsample be sure to do it using evenly divisible numbers, 44.1 to 88.2 to 176.4 and 96 to 192, etc. This generally sounds better than when the numbers are odd.
Scotty
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: richidoo on January 25, 2018, 01:06:28 PM
No upsampling for me. No matter what kind of processing, I can always hear it, and I never prefer it.
Even Roon's DSP engine takes away more than the upsampling adds, imo.
I hear it as a dulling of the harmonic complexity in high frequencies. It removes the fun part of the sound that I pay so much for high end equipment to be able to hear.
I've had the same experience with Cary upsampling CD player, Tact DSP preamp, JRiver to 192k, Roon to 384k and even DSD,  upsampling sounds similar on all of these to me. It's not the hardware, it's the math.
I just prefer the original master resolution.

SoX upsampling improved the SQ of Moode Audio on RPi, but SoX did not make up for the frustration of the Moode bugs. Volumio no upsampling was needed to sound good.

IMO there's far more to be gained by attenuating jitter from digital playback than messing with the waveform math.
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: Triode Pete on January 25, 2018, 02:02:13 PM
Would be interesting to hear from Dave and Pete as they are NOS guys that use a Mac mini. What is their opinion of DSD 512 and do they think upsampling has any redeeming qualities.
Nick

As always, it depends on the recording! But, I've found the majority of decent recordings that "No Upsampling" sounds best! It's amazing the analogue sound you can extract from decent Redbook recordings...

To me, Upsampling adds that really hi-fi'sh sound that slowly gives me digiitus!

My $0.02,
Pete
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: P.I. on January 26, 2018, 10:57:03 AM
Everything that I have to say has to be put through the BS filter of a 70 year old, so YMMV.

I have several DACs of varying capabilities.  I do not have a DSD capable DAC and have no desire to buy one.  I have a ton of data almost all of which is in Redbook format.  There are so few (relatively) files available in true hi-rez that appeal to me.  I live for the music, however it is presented.  I have heard all of these high zoot systems with dacs that cost more than a car and always come away "meh".

I am a NOS guy and will be until I hear one that meets my performance/cost criteria.

There is just something about NOS done right that appeals to me and it falls into the PRAT realm.  I can't hear DC to light and hear music as primarily a midrange event.  Sure, the bottom octave or two is where a lot of meat lives, but impact is a higher harmonic order happening.  Bass guitar articulation and tunefullness lies in the 1K - 2K range along with kick drum smack.  I'll always take clean, pristine, bell like highs over what a lot of detail freaks (come on - you know who you are  :lol: ) prefer in the top octave, some of which I can still hear thankfully.  As a drummer (rock and roll concussionist) I always give a intense listen to transient impact, skin tone and harmonic series of cymbals.  If these aren't right then I'm out.  Same goes for well recorded vocals.  Breathiness, sibilants, plosives and fricatives need to sound natural, not bigger than life.  Too many of the hi-rez DACs and systems that I have heard are technicolor instead of natural.

More is nice, but better is better for me.

I'm crappy at reducing emotion down to verbage.   :?
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: dflee on January 26, 2018, 01:16:41 PM
So I've got a cdp and an upsampling dac.
Am I reading this correctly that I should get something for jitter and leave it at that. What should I get if I just use the cdp without bypassing the internal dac for better sound?

Don
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: richidoo on January 26, 2018, 01:57:58 PM
No shoulds here... ;)  but you can reduce jitter on the CDP's digital output with iFi purifier 1 (SPDIF.) I have one, it works good, jitter is only 60pS.

If you use the CDP analog outputs (without an external DAC) then there's no easy fix for the jitter. Surgery is necessary to upgrade (http://www.tentlabs.com/Products/cdupgrade/xo2xo3/) the CDP's word clock.
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: Nick B on January 26, 2018, 06:19:28 PM
Interesting reading for sure. I confess to being a detail freak. But I’m trying to change... really I am!!
Hence I’m going the tubey, more musical route. As an aside, I sold two big ticket items, so i can spend some money. Yay 👍
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: mfsoa on January 26, 2018, 07:06:08 PM
I have 2 computers hooked up to my dac (down from 3 - didn't use Moode very much on the Pi). I consider having the option to alter the digital filtering essential to maximizing the sound quality of one's system. It's cool too. Haven't dabbled quite yet in deconvolution but I'm sure that's on the way.

- Pi with an Allo Digione, running PiCorePlayer, coax spdif to dac. PiCorePlayer has the ability alter the type of filter (linear phase, minimum phase, everywhere between), filter slopes, upsampling etc. The ability to shift between linear phase and minimum phase  results in a huge change in sound quality (on audiophile terms). Much more than a cable swap, linear power supply etc. If you can't alter this in your current setups then you are stuck with what the designer picked. All digital filters are wrong. I like being able to select the least-wrong one for my tastes. I like the sound of everything upsampled to the max the spdif can take, 176 or 192, even multiples, with a filter that's half way between linear and minimum phase.  Not sure if the iFi Purifier will improve the Digione - would like to try.

- Minix Neo Z83-4, Win 10, fanless, cute. This runs Jriver, Tidal and is an HQPlayer NAA device, so nearly unlimited ability to choose the digital mangling I want. I prefer that to having someone else choose it for me and hope that their choice always matches the one I would have made for every piece of music for all time... Upsampling to 4X DSD is interesting. It always sounds better on acoustic music, but I often prefer plain old PCM for more electronic stuff. The output of this PC goes to an IfI nano IUSB3.0. In HQPlayer and JRiver I usually upsample everybody to the 356/384 family.


So then it goes onto a dac  (Gustard A20H, longing for a Brooklyn DAC+) which further upsamples again, so I really have no freaking idea what the dac thinks when it sees data that has already been upsampled by a computer. The theory is that the less work the DAC has to do t he better, so have your high-power PC do as much of the heavy lifting as possible.

So, to upsample or not to upsample, that is the question.  It's just so cheap and easy to get so much capability - I'd feel naked without it.
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: dBe on January 26, 2018, 10:17:34 PM
Had a weird couple of days where I'm away from work, so I've had some time to think about this a 'bit'.

My choice for gear in high-ish end audio always gravitates to very simple, well executed electronics.  I prefer triode gear for amplification built with exemplary parts.  Every time an audio signal encounters a input connector, a piece of wire, a solder joint, resistors or (gasp!) capacitors, output connectors the signal is altered by non-linear distortions.  Take a piece of SS gear and the distortions are further altered by multiple devices with tiny internal traces made of whatever, more and more gain stages, FIFO buffers, gate delay, latency and much more.  All of these things screw with phase by flipping absolute phase and phase angle rotations caused by every component in the signal chain.  These changes and distortions are relatively innocuous in analog reproduction because there is much less impact in timing distortion and phase.  When absolute phase is screwed with in analog, it is much easier to benignly herd the audio signal back into line.

Digital sins are several orders of magnitude greater than their analog, linear phase brethren.  By definition the digital end of our chosen playback change can be really hard to quantify differences when our preselected.

Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: rollo on January 29, 2018, 09:07:18 AM
Had a weird couple of days where I'm away from work, so I've had some time to think about this a 'bit'.

My choice for gear in high-ish end audio always gravitates to very simple, well executed electronics.  I prefer triode gear for amplification built with exemplary parts.  Every time an audio signal encounters a input connector, a piece of wire, a solder joint, resistors or (gasp!) capacitors, output connectors the signal is altered by non-linear distortions.  Take a piece of SS gear and the distortions are further altered by multiple devices with tiny internal traces made of whatever, more and more gain stages, FIFO buffers, gate delay, latency and much more.  All of these things screw with phase by flipping absolute phase and phase angle rotations caused by every component in the signal chain.  These changes and distortions are relatively innocuous in analog reproduction because there is much less impact in timing distortion and phase.  When absolute phase is screwed with in analog, it is much easier to benignly herd the audio signal back into line.

Digital sins are several orders of magnitude greater than their analog, linear phase brethren.  By definition the digital end of our chosen playback change can be really hard to quantify differences when our preselected.

  An invert polarity switch is a good thing to have. Besides what components may do the recordings themselves can be the out of phase culprit.
When you hear 44.1 done right with no chip set and a dedicated transport you will realize the difference missing from upsampling, chip DAC's. So easy to hear.


charles
Title: Re: REDBOOK To Upsample or not is The Question Is it Noble ?
Post by: dBe on January 29, 2018, 10:05:57 PM
Had a weird couple of days where I'm away from work, so I've had some time to think about this a 'bit'.

My choice for gear in high-ish end audio always gravitates to very simple, well executed electronics.  I prefer triode gear for amplification built with exemplary parts.  Every time an audio signal encounters a input connector, a piece of wire, a solder joint, resistors or (gasp!) capacitors, output connectors the signal is altered by non-linear distortions.  Take a piece of SS gear and the distortions are further altered by multiple devices with tiny internal traces made of whatever, more and more gain stages, FIFO buffers, gate delay, latency and much more.  All of these things screw with phase by flipping absolute phase and phase angle rotations caused by every component in the signal chain.  These changes and distortions are relatively innocuous in analog reproduction because there is much less impact in timing distortion and phase.  When absolute phase is screwed with in analog, it is much easier to benignly herd the audio signal back into line.

Digital sins are several orders of magnitude greater than their analog, linear phase brethren.  By definition the digital end of our chosen playback change can be really hard to quantify differences when our preselected.

  An invert polarity switch is a good thing to have. Besides what components may do the recordings themselves can be the out of phase culprit.
When you hear 44.1 done right with no chip set and a dedicated transport you will realize the difference missing from upsampling, chip DAC's. So easy to hear.


charles
Polarity has a very real influence upon the audio that we listen to and it is relatively easy to hear.  It is also pretty simple to implement in most systems.  Sometimes it is as simple as switching output phase to a preamp from the DAC or analog source.

One of my gurus wrote an interesting paper and presented it to the AES, but unfortunately got poo-pooed by a lot of "scientists" and his work largely has gone unheeded.

Think about it this way.  With absolute phase observed at the speaker output, dynamics will be honestly served.  Pure physics:  with, say, a big drum hit the speaker is going to push a compressible gas (fluid) and portray the transient as intended- a fast outgoing waveform.  If the speaker is operating anti-phase, the diaphragm is going to try to suck an equal amount of air compared to the positive going pulse.  Ain't gonna happen.  Air is to a point elastic in that it is more efficient to push it than pull it.

Read Clark's thoughts here:  https://positive-feedback.com/Issue1/cjwoodeffect.htm

Food for thought that will go a long way in the consideration of PRAT and why it is some of the magic that we hear as 'right'.