Chords New M -Scaler

[austinpop]

27 minutes ago, rayl1234 said:

all other chatter is just noise until this situation changes.

 

I couldn't agree more. 

 

I am a  fan of, and own a copy of HQPlayer, but I think it is unseemly for competitors like @Miska to come here and sow FUD about the Chord product. We won't know if "HQPlayer +  the best HQP-friendly DAC (T+A DAC 8 DSD?)" sounds better than the "Chord Hugo M-Scaler + Chord Hugo TT 2" until people compare them when the time comes and post their impressions.

 

All this arguing about the math and the computation - FPGA vs. general-purpose CPUs and GPUs - is just noise. 

 

Sadly, this chatter will continue unabated until the product is available. Ultimately, the proof of the pudding (dessert, for the Americans) is in the eating (eating, for the Americans).

 

If I had my druthers, I'd declare all the HQPlayer discussion here as OT for this thread. Folks who want to continue that discussion should either take it back to the HQPlayer thread, or open a new thread.

[auricgoldfinger]

23 minutes ago, austinpop said:

All this arguing about the math and the computation - FPGA vs. general-purpose CPUs and GPUs - is just noise. 

 

If I had my druthers, I'd declare all the HQPlayer discussion here as OT for this thread. Folks who want to continue that discussion should either take it back to the HQPlayer thread, or open a new thread. 

 

I agree completely.  Thank you for speaking up.

[seeteeyou]

On 7/27/2018 at 11:21 PM, austinpop said:

So bottom line, while you could get HMS upsampling to 192kHz to most DACs, that only gives you access to 1/4 million taps. Should a DAC be capable of 384kHz on its S/PDIF input, that would give you 1/2 million taps.

 

ComTrue made something called CT7601 Audio USB Bridge for 200 bucks and at the very least it's able to record 0.5M taps via that single S/PDIF output @ 384kHz

 

https://www.head-fi.org/threads/review-comparison-of-5-high-end-digital-music-servers-aurender-n10-cad-cat-server-totaldac-d1-server-auralic-aries-audiophile-vortex-box.787020/page-79#post-14119139

In other words, the physical presence of M Scaler should be no longer required after the recording is done. In other words, we could split the costs of acquiring that, we could rent that, we could borrow that from a buddy, we could pay an owner to do the recording, or owners could even sell that after an entire music library is done, so on and so forth.

 

Quote

It uses the mono data format on AES3, in order to double the sample rate, so this is an industry standard. But I don't know any DACs that use it today.

 

Since the coaxial input of CT7601 actually supports up to 768kHz, maybe we could use two of them to record each mono channel separately / simultaneously and then merge them into stereo *.WAV files afterwards? Editing *.WAV files should be relatively easy since PCM shouldn't be a challenge by any means.

[Adyc]

3 hours ago, auricgoldfinger said:

 

I agree completely.  Thank you for speaking up.

 

+1

[Jud]

4 hours ago, rayl1234 said:

 

I read it differently.  I read it as the particular set of coefficients that achieve a certain error bound but are computable on the particular FPGA's capacity -- computable in the sense that unlike math, you don't have infinite precision...  The term "better than 16 bits" is thrown around, but unfortunately, that is one of the areas where I've not gotten a clear answer.  Is the implication that:

 

1. Each coefficient is w/in 2^-16 of the coefficient of the mathematical 1MM windowed sinc?  (This would say very little interesting as certain coefficients accumulate much greater errors than others, obviously)

2. Compared to using a mathematical 1MM windowed sinc, the results computed will match the mathematical computation when quantized to 16 bits?

3. For all bandwidth limited signals, the result of applying the filter will result in the equivalent digital values as a 705.6kHz sampling of the analog signal. (This would be the strongest defition, but this has not been confirmed much less had a proof presented).

 

However, I think my original thesis still stands -- although technically "feasible" in software, I haven't seen a truly packaged solution even for offline conversion, much less for streaming. (I will say, not to bring up a big debate on the topic, that I have my reasons to be a bit skeptical of "5%" CPU bec my admittedly unoptimized offline test code when I tried it last fall was closer to 60% of a 4 core CPU... Of course, on a GPU with 2000+ DSP cores, it would be easy, but would probably consume around 70+ watts of power.)  Thus, if you like what WTA does for you (and I do), all other chatter is just noise until this situation changes.

 

I can't see how CPU usage for offline conversion is relevant to anything.

[Jud]

If anyone wants to see whether using what I will technically call a "shitload" of taps will make a difference, you can try Audirvana Plus with the iZotope SRC, which does anywhere from 10,000 to 2,000,000 taps. Or you can get iZotope itself on trial and do the conversion offline if you like.

[adamdea]

For mathematicians out there- one of the things which has been puzzling me is that when I have played with using sinc function as a filter, it’s apparent that the windowing function seems to matter more than the width of the window to the frequency response (certainly comparing sensible window short sinc with long rectangular window)  

 

Would a bespoke window function be more promising than increasing of tap length to increase accuracy of reconstruction? Or perhaps an increase in precision of output? 

[adamdea]

7 hours ago, austinpop said:

 

I couldn't agree more. 

 

I am a  fan of, and own a copy of HQPlayer, but I think it is unseemly for competitors like @Miska to come here and sow FUD about the Chord product. We won't know if "HQPlayer +  the best HQP-friendly DAC (T+A DAC 8 DSD?)" sounds better than the "Chord Hugo M-Scaler + Chord Hugo TT 2" until people compare them when the time comes and post their impressions.

 

All this arguing about the math and the computation - FPGA vs. general-purpose CPUs and GPUs - is just noise. 

 

Sadly, this chatter will continue unabated until the product is available. Ultimately, the proof of the pudding (dessert, for the Americans) is in the eating (eating, for the Americans).

 

If I had my druthers, I'd declare all the HQPlayer discussion here as OT for this thread. Folks who want to continue that discussion should either take it back to the HQPlayer thread, or open a new thread.

I disagree. I may be eccentric but I’m more interested in what dac/ dsp designers have to say for themselves than in the latest twists of consumers’ purchasing/reward compulsion. 

This stuff is basically fantasy maths/engineering. Not imaginary, but way beyond thresholds of hearing. So one is buying into technical accomplishment as an end in itself. One might as well listen to what people have to say. I’m not interested in cults- so I’m not interested in just rob watts’ say so. 

 

Miska has a point about whether watts and other people’s designs actually do reduce modulation and hf spuria- and I’m happy to listen- up to a point. 

 

Obviously that’s not an excuse for pure shilling,  or the incomprehensible gibberish  and tedious repetition you get from some other posters in this forum. But if RW wants to set him right all the better. 

[guiltyboxswapper]

8 hours ago, austinpop said:

 

I couldn't agree more. 

 

I am a  fan of, and own a copy of HQPlayer, but I think it is unseemly for competitors like @Miska to come here and sow FUD about the Chord product. We won't know if "HQPlayer +  the best HQP-friendly DAC (T+A DAC 8 DSD?)" sounds better than the "Chord Hugo M-Scaler + Chord Hugo TT 2" until people compare them when the time comes and post their impressions.

 

All this arguing about the math and the computation - FPGA vs. general-purpose CPUs and GPUs - is just noise. 

 

Sadly, this chatter will continue unabated until the product is available. Ultimately, the proof of the pudding (dessert, for the Americans) is in the eating (eating, for the Americans).

 

If I had my druthers, I'd declare all the HQPlayer discussion here as OT for this thread. Folks who want to continue that discussion should either take it back to the HQPlayer thread, or open a new thread.

 

Again, worth me repeating this.  TT2 by Watt's own words isn't superior to the DAVE.  Had the DAVE around twice (as I was getting fed up of HQPlayer + ASIO driver recovery issues and running a PC), and it was clearly outshone by HQPlayer and the T+A.  Especially the treble energy matter, that's what pushed it over the edge for me infact.

 

Again, I wanted a simple "turn it on - music comes out" box, desperately, so was happy to keep the DAVE had it kept up.  Was wonderful having a single box solution, takes any input etc.

 

I actually think its worth while having the other options and how they achieve their outcomes in this thread.  

 

Understanding the method behind the outcome is quite interesting in itself; often most conversations hifi related circle around more basic matters...

[Whitigir]

4 hours ago, Jud said:

If anyone wants to see whether using what I will technically call a "shitload" of taps will make a difference, you can try Audirvana Plus with the iZotope SRC, which does anywhere from 10,000 to 2,000,000 taps. Or you can get iZotope itself on trial and do the conversion offline if you like.

Which iZotope? And does it use WAT filter ? Or Linear phase

[Miska]

10 hours ago, austinpop said:

I am a  fan of, and own a copy of HQPlayer, but I think it is unseemly for competitors like @Miska to come here and sow FUD about the Chord product. We won't know if "HQPlayer +  the best HQP-friendly DAC (T+A DAC 8 DSD?)" sounds better than the "Chord Hugo M-Scaler + Chord Hugo TT 2" until people compare them when the time comes and post their impressions.

 

Well, I don't know where I'm putting out some FUD, instead of responding to FUD spread by Rob Watts about software implementations and general purpose CPU/GPU (sometimes echoed on this forum too).

 

10 hours ago, austinpop said:

All this arguing about the math and the computation - FPGA vs. general-purpose CPUs and GPUs - is just noise.

 

I think this is perfectly good topic for discussion, whether this is the correct thread for that is another question. I think that is one of the most interesting topics to discuss.

 

10 hours ago, austinpop said:

If I had my druthers, I'd declare all the HQPlayer discussion here as OT for this thread. Folks who want to continue that discussion should either take it back to the HQPlayer thread, or open a new thread.

 

I agree, so I'd rather keep the discussion more on generic software vs firmware level.

[Miska]

11 hours ago, rayl1234 said:

However, I think my original thesis still stands -- although technically "feasible" in software, I haven't seen a truly packaged solution even for offline conversion, much less for streaming. (I will say, not to bring up a big debate on the topic, that I have my reasons to be a bit skeptical of "5%" CPU bec my admittedly unoptimized offline test code when I tried it last fall was closer to 60% of a 4 core CPU... Of course, on a GPU with 2000+ DSP cores, it would be easy, but would probably consume around 70+ watts of power.)  Thus, if you like what WTA does for you (and I do), all other chatter is just noise until this situation changes.

 

Mind you, typical digital room correction filter varies from 64k taps to 256k taps at 44.1 kHz sampling rate. Adapted to 705.6 kHz that is is 1M taps to 4M taps. And computer running such filters doesn't even break a sweat while also doing upsampling to DSD rates at the same time.

 

That's why I find it a bit funny that running a million tap filter would be somehow big technical achievement in itself. Maybe it is for those cheaper FPGA models, probably less so for the bigger FPGA's like UltraSCALE Virtex/Kintex models, but I haven't seen anybody using those for audio yet.

 

[jabbr]

11 hours ago, austinpop said:

All this arguing about the math and the computation - FPGA vs. general-purpose CPUs and GPUs - is just noise. 

 

Sadly, this chatter will continue unabated until the product is available. Ultimately, the proof of the pudding (dessert, for the Americans) is in the eating (eating, for the Americans).

 

If I had my druthers, I'd declare all the HQPlayer discussion here as OT for this thread. Folks who want to continue that discussion should either take it back to the HQPlayer thread, or open a new thread.

Let’s see ... this thread is about an unreleased product. 

 

1) Claims that it beats HQPlayer

2) Discussion about how it works

 

(summary of above posts)

 

So sure let’s not compare to other products and then they’d not get mentioned. 

 

But no technical discussions including theories? Hmm that leaves fanboi! Yeah rah rah!

 

When there no pudding just a recipe, the discussions about the recipe. Or not.

[rayl1234]

10 minutes ago, jabbr said:

Let’s see ... this thread is about an unreleased product. 

 

1) Claims that it beats HQPlayer

2) Discussion about how it works

 

(summary of above posts)

 

So sure let’s not compare to other products and then they’d not get mentioned. 

 

But no technical discussions including theories? Hmm that leaves fanboi! Yeah rah rah!

 

When there no pudding just a recipe, the discussions about the recipe. Or not.

 

I will differ....

 

While Hugo mscaler is unreleased, Blu2 mscaler has been in purchasers’ hands since spring 2017....

 

The pudding has been delivered, just not a new dish for it. 

 

And on miska’s comment, there is a difference between 1mm for 1 bit vs PCM, but that isn’t really the point.

 

I had tried some (not all or even many) hqp settings, albeit maybe not in combo with the best dac for hqp... but that’s a big part of what made me end up dumping my entire dac chain for bku2+Dave. Of course I tried a bunch of products... lampi, dcs... heck, it was a big purchase.  But Chord was a packaged solution that worked great (very natural/at ease would be my best short phrase) out of the box for all my use cases, even with EDM tracks, not just orchestral. (Not surprising as I recall RW mentioning listening to quite a bit of edm.)

 

That’s my bottom line. This thread suggested that I may have missed an alternative, but short of building quite a few pieces (on the hqp thread), I do not believe that to be the case. If so, great—I can upgrade my office rig. Would still have no regret over the chord experience. 

[jabbr]

1 minute ago, rayl1234 said:

This thread suggested that I may have missed an alternative, but short of building quite a few pieces (on the hqp thread), I do not believe that to be the case. If so, great—I can upgrade my office rig. Would still have no regret over the chord experience. 

 

No problem. I’ve said above what I think. It’s just that if the comparison to HQPlayer is made, then it invites a whole discussion. I can’t see how @Miska should be faulted for participating — he didn’t barge in — rather commented on statements made about his product. 

Also when folks bring up technical details it invites a technical discussion — which is totally fine. Particularly since the product is not released. I’ve no doubt the whole Chord stack sounds great together.

[Whitigir]

From what I experiences and the stands of debates from Chord.  Their products is within it’s own Eco system.  I would never accept the fact that USB sounds better than I2S.  The native language of DAC is I2S, and USB needs interpreter.  Just like anything else, interpreter is never original, period.

 

The price for Chord products are expensive, and whoever believes it is best, they can always enjoy it.

 

After all, satisfaction is the goal.  It could be objective such as brand and pricing, or subjective of sound performances 

[mansr]

3 hours ago, Whitigir said:

Which iZotope? And does it use WAT filter ? Or Linear phase

Assuming you meant WTA, it appears to be nothing more than a made-up name for a windowed sinc function. Extracting the exact filter coefficients from the M Scaler would be trivial.

[Jud]

3 hours ago, Whitigir said:

Which iZotope? And does it use WAT filter ? Or Linear phase

 

Either version of iZotope.

 

IZotope does allow you to configure the filter to be linear phase, yes.  That's a pretty routine thing.

 

Haven't read the thread closely enough to figure out what a WAT filter is (with the name similarity, I suppose it's the filter Chord uses), so I can't say how one might configure the Chord filter to be similar.

 

The main point was that if you wanted to see what a particular aspect of the Chord filter - use of lots of taps - might do for you, this would be an opportunity to try that particular aspect for free. 

[mav52]

2 hours ago, Miska said:

I think this is perfectly good topic for discussion, whether this is the correct thread for that is another question. I think that is one of the most interesting topics to discuss.

 

It all depends on how far the discussion moves towards your product without a deeper move into the chord product  which is difficult unless Chord joins the discussion  .  So it at this time is pretty one sided.

[ecwl]

28 minutes ago, mansr said:

Assuming you meant WTA, it appears to be nothing more than a made-up name for a windowed sinc function. Extracting the exact filter coefficients from the M Scaler would be trivial.

Yes. We can debate whether it’s ethical to steal someone else’s filter coefficients even though it’s just math and a windowed sinc function so it doesn’t have a copyright on it. I’m quite ambivalent on the matter.

Since Chord Blu Mk 2 has been out for almost 1-1/2 years, if someone wants to figure out the coefficients, people could have figured out the coefficients by now (even though there is some minimal copy-protection around that as in Blu2 blocks some signals which can generate the coefficient directly).

I think in some ways this forum discussion has been helpful because we have heard from a number of people why they wouldn’t bother to figure out these coefficients because:

1) people don’t think the WTA coefficients are better than the filters they can create on their own (although I have yet to hear a filter that sounds better than Blu2’s or even Mojo’s but one can argue may be it’s just my personal preference)

2) people don’t think you need the long tap lengths/all these coefficients and there are other ways to generate more optimized, more computationally efficient filters (although once again I have yet to hear a filter that I like better than Blu2 or Mojo’s)

3) people don’t think 1fs to 16fs is the right approach as a start, e.g. why not go straight from 1fs to DSD512 (and we have heard some people liking that approach better here but there are also lots o people who prefer DAVE +/- Blu2 in the Head-Fi forum over the 1fs to DSD512 software + hardware approach)

 

I do think once Hugo M-Scaler comes out, somebody would eventually publish the WTA filter coefficients. We can then see what kind of computational power is required to do these calculations. Rob Watts recently commented that he doesn’t think a generic CPU can do them in real-time but he definitely thinks a CPU+GPU can and he actually suggested that people should just go ahead and do it in the Head-Fi forum. But obviously Rob Watts is not going to publish his WTA filter coefficients. Just like nobody who charges for their software upsampler so far has been willing to open source their code so that we can view them.

 

 

 

[mansr]

12 minutes ago, ecwl said:

Yes. We can debate whether it’s ethical to steal someone else’s filter coefficients even though it’s just math and a windowed sinc function so it doesn’t have a copyright on it. I’m quite ambivalent on the matter.

The point is that it's just another windowed sinc (JAWS) filter. It's not even clear that this is the best choice for audio applications.

[ElviaCaprice]

11 minutes ago, ecwl said:

I do think once Hugo M-Scaler comes out, somebody would eventually publish the WTA filter coefficients. We can then see what kind of computational power is required to do these calculations. Rob Watts recently commented that he doesn’t think a generic CPU can do them in real-time but he definitely thinks a CPU+GPU can and he actually suggested that people should just go ahead and do it in the Head-Fi forum. But obviously Rob Watts is not going to publish his WTA filter coefficients. Just like nobody who charges for their software upsampler so far has been willing to open source their code so that we can view them.

 

 

Agree, would be a fun exercise to try.  But I have a hunch that Rob Watts knows something about his design that would make it superior in SQ to any CPU+GPU  design.  Regardless of all the FUD given here, I have yet to hear of a step by step procedure that produces an equivalent SQ result as the BLU MKII or future Hugo M-Scaler.  Which only confirms the findings of Romaz and others that have tried with DAVE using HQP.

 

As much as I would like a software solution, low power, it remains elusive.  So I look forward to the hardware solution of Hugo M-Scaler.

[auricgoldfinger]

Amidst all the FUD, I am surprised that no one has brought up the fact the HQP hangs and must be force closed almost every time the sample rate of a track changes.  While this issue is manageable when playing entire albums, it precludes the use of randomized play lists.  Oh, and there's also the wonderful HQP user interface.

 

Just sayin' since the thread is already so off topic.

 

[adamdea]

1 hour ago, mansr said:

The point is that it's just another windowed sinc (JAWS) filter. It's not even clear that this is the best choice for audio applications.

Could you expand on your thoughts on the best choice. -assuming we are sticking with linear phase is this just a question of the right windowing function or do you mean that we shouldn’t bother with an actual sinc?  Am I right in thinking that you can make something with more attenuation for a given filter length by not using a windowed sinc?

there is a plausible sounding theory out there that (to be on the safe side)  linear phase filters should not be longer than something to do with the ears’ filter bins possibly the time constant. I think that Jj over at HA and Fokus have both mentioned this point, but I have never quite grasped it. 

[austinpop]

 

1 hour ago, mav52 said:

which is difficult unless Chord joins the discussion  .  So it at this time is pretty one sided.

 

3 hours ago, Miska said:

 

Well, I don't know where I'm putting out some FUD, instead of responding to FUD spread by Rob Watts about software implementations and general purpose CPU/GPU (sometimes echoed on this forum too).

 

 

I think this is perfectly good topic for discussion, whether this is the correct thread for that is another question. I think that is one of the most interesting topics to discuss.

 

 

I agree, so I'd rather keep the discussion more on generic software vs firmware level.

 

I have no problem with 2 designers, with proven solutions under their belt, having a debate. If RW were here, and it was just Miska and him debating, I would welcome it and follow with great interest. This may not be your view, but speaking as a "neutral," I find it fascinating that you guys, while being on opposite ends of the design spectrum, have clearly been able to demonstrate outstanding sound quality with your approaches.

 

However, that's not what's happening here. It's a one-sided debate between you, someone with deep knowledge and skill, who's actually delivered a product, and people with opinions. No good usually comes from that.  That's why I am suggesting that unless RW wants to join this debate, it may be best for you to refrain from criticism, at least here in this thread, since it will inevitably come across as FUD.