Chords New M -Scaler

[mansr]

7 hours ago, ecwl said:

But if a Chord DAC owner can identify a 1fs to 16fs software filter that improves on the Chord DAC’s own WTA filter, I’d be more than happy to give it a listen for myself and to compare that to the M-Scaler in Blu2.

Can you play 768 kHz files with a software player? If I prepared some test files, would you give them a listen?

[ecwl]

4 minutes ago, mansr said:

Can you play 768 kHz files with a software player? If I prepared some test files, would you give them a listen?

Sure. Just PM me with the link or if you want to share with everyone, post the link here. 

 

Listening using Mojo is easier. Blu2/DAVE is setup mainly for Roon. Might require more effort to test that out.

[mansr]

7 minutes ago, ecwl said:

Sure. Just PM me with the link or if you want to share with everyone, post the link here. 

 

Listening using Mojo is easier. Blu2/DAVE is setup mainly for Roon. Might require more effort to test that out.

What source format would be best for this test? CD quality or something else? Any particular style of music? How long a sample do you need? I don't have a problem providing a full track, but an excerpt may be preferable from a copyright point of view. Or you could give me a sample you think will be revealing.

 

Whatever the source, I'll upsample it 16x using a few different filters. Then you, or anyone else, can compare them to the Chord and report your impressions.

[ecwl]

 

17 minutes ago, Whitigir said:

However, if you claim that you rather sell people equipments that Down Convert Native DSD from a Studio to PCM regardless of rate, because you believe that your other equipments will do a more accurate jobs ? 

 

Are you kidding me ? Do you know how much a Studio invest in their resources ? To be outdone by your mere Chord Stack ? Give me a break

 

Hmmm... I can see you’re very passionate about this

Except the only thing that is truly native to sound is the original analog sound wave. Whether it is PCM or DSD recordings, they are all digital sampling with noise shaping.

Moreover, a native DSD recording is not really played back natively in the way you’d like to think nowadays.

Let’s say it’s silence so the signal is 0101010101010101... at 2.822MHz.

Let’s take the DSC1 open source design which has 32-elements so it can generate anywhere between 0 to 32. When you play “natively” on this “native” DSD design, the DAC is not oscillating between 0 and 32 back and forth. It actually starts with 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, ..., 16, 16, 16, 16, 16, 16, 16, 16 ,16, 16, 16, 16, 16, 16, 16, ... at 2.822MHz let’s say. (Or at least that’s my understanding of it. People with more expertise can correct me if I’m wrong). 

Yes, I know conceptually, that’s still quite far from converting DSD to PCM then to a different multi-bit SDM presentation. But the principle of “native” is a relative concept.

As to your comment that Studios invested lots of resources on this so they must know how to do a better job. By that principle, SpaceX can’t launch a rocket better and cheaper than NASA?

 

At the end of the day, we like our music the way we like them. If you like your DSD tracks and you like them played through your favorite DSD DAC, that’s fantastic. It’s your hard-earned money. You should pay for the DAC you like the most and enjoy, like the rest of us.

[auricgoldfinger]

1 hour ago, Miska said:

Last time I checked they had just 120 dB stop-band attenuation, which in my opinion is not enough, or in other words I don't see reason why not do more. That's why I use 192 dB so that it matches at least 32-bit resolution... But anyway, I think there are so many aspects in a filter design, that single parameter like number of taps is pointless without all the other information...

 

 

Since your software seemingly can do no wrong, when do you plan to address the fact that HQP hangs and must be force closed almost every time the sample rate of a track changes?  Oh, and there's also the wonderful HQP user interface that causes most users to have to buy a Roon subscription.

 

As someone who has used HQP for 1.5+ years, I'd say it's not exactly ready for prime time.  Maybe it's time to stop taking pot shots at Chord on the M-Scaler thread and focus your valuable time, energy, and experience on your own product.

 

 

[Whitigir]

I just take that DSD from a Studio is processed by the Studio and engineers.  SpaceX resources is different than NASA.  Both of them have engineers and a huge team behind it to keep tinkering with different goals in mind to get the result.  You honestly think NASA can not do it like SpaceX ? They just don’t want to (thanks to all that budget cutting)

 

Back at Native DSD.  It is modulated and processed with all the digital filters, and whatever else in the chain.  Yes, I am aware that different DAC will alter the sound signatures.  But my point stands, Native DSD from an SACD is what the engineers, the artist, and the Studio believe and want the consumer to hear.

 

I have 0 problems with down converting your Native DSD contents and have it played back by S/D system, as I said that I enjoy both DSD and PCM in different circumstances 

 

I do have problems when someone say that they believe they have perfected a software and hardware that put other studios and engineers to shame

[ecwl]

3 minutes ago, mansr said:

What source format would be best for this test? CD quality or something else? Any particular style of music? How long a sample do you need? I don't have a problem providing a full track, but an excerpt may be preferable from a copyright point of view. Or you could give me a sample you think will be revealing.

 

Whatever the source, I'll upsample it 16x using a few different filters. Then you, or anyone else, can compare them to the Chord and report your impressions.

I suspect the best source format would be CD and then upsampled to 16x. As you’ve noticed my comparison tracks, I find it’s easiest to hear the differences between different tap length WTA filters on Chord DACs either with specific instruments with obviously timbre and note strikes, e.g. piano music, or with specific instruments that highlights transients/timing, e.g. guitar plucks, hand clapping, finger snapping, drum strikes, cymbals. That is not to say that everything else don’t sound better with longer tap length WTA filters, it’s just that these things are easier to hear for comparison purposes. Honestly, if you record yourself playing the piano or clapping your hands or striking drums, or doing your own rendition of Cups, even that can work. And now that I think about it, it doesn’t even matter whether it’s 44.1kHz source or 48kHz. Since I don’t fully understand many aspects of filtering, I’m also wondering if it’s possible if you record something at 768kHz and then downsample it to 48kHz, and then run the various filters and then compare the mathematical delta between the upsampled files to the original 768kHz. Although I presume that’ll be a bit meaningless since the result would be more a function of the antialiasing setting? Not sure. I should stop this stream of consciousness. Thanks for providing some tracks.

[mansr]

5 minutes ago, ecwl said:

As you’ve noticed my comparison tracks, I find it’s easiest to hear the differences between different tap length WTA filters on Chord DACs either with specific instruments with obviously timbre and note strikes, e.g. piano music, or with specific instruments that highlights transients/timing, e.g. guitar plucks, hand clapping, finger snapping, drum strikes, cymbals.

So some solo guitar would be good?

[ecwl]

2 minutes ago, mansr said:

So some solo guitar would be good?

Sure.

[mansr]

20 minutes ago, ecwl said:

Let’s take the DSC1 open source design which has 32-elements so it can generate anywhere between 0 to 32. When you play “natively” on this “native” DSD design, the DAC is not oscillating between 0 and 32 back and forth. It actually starts with 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, ..., 16, 16, 16, 16, 16, 16, 16, 16 ,16, 16, 16, 16, 16, 16, 16, ... at 2.822MHz let’s say. (Or at least that’s my understanding of it. People with more expertise can correct me if I’m wrong).

Yes, I know conceptually, that’s still quite far from converting DSD to PCM then to a different multi-bit SDM presentation. But the principle of “native” is a relative concept.

The DSC1 is a 32-sample moving average filter followed by the D/A conversion proper. This provides an initial digital low-pass filter which eases the requirements of the analogue filter. I see nothing wrong with such an approach.

[rayl1234]

53 minutes ago, Whitigir said:

Everybody has the right to be proud of their products.  However, the way the M-Scaler is doing and RW marketing, it offend other people, including the engineers, the artists, the studios, and even enthusiast 

 

see ? SACD from a Native Studio meant that it is a perfected tracks in everything that the studio want the listener to listen to, without tinkering and alternating the sound signatures.  That is what Native DSD is all about

 

Now, let’s not get into why DSD is better or worse than PCM.  I enjoy both under different circumstances.  However, if you claim that you rather sell people equipments that Down Convert Native DSD from a Studio to PCM regardless of rate, because you believe that your other equipments will do a more accurate jobs ? 

 

Are you kidding me ? Do you know how much a Studio invest in their resources ? To be outdone by your mere Chord Stack ? Give me a break

 

I am not flaming on Chord, but I am not a blind sheep either.  I used to respect Chord, but losing it every day.

 

I wouldn’t feel so bad if Chord simply states “by using M-Scaler, the consumers can have their own Chord house sound even from Native DSD tracks”.  This statements is different than “recovering the most accurate Informations at ADC stages and reconstruct it more accurately”.  Huge different 

 

 

I will admit that I have not conducted an evaluation of DSD content so cannot comment.  I will also admit that in choosing Chord, I was well aware that DSD may be a weakness (and certain MQA *is* a weakness).... and the choice was made in light of the fact that I haven't seen any value in MQA after having tried it with the Brooklyn for a few months, and I had little DSD content.

 

DSD is a complicated topic bec we live in an unfortunate world. DSD was invented to get consumers to buy reissues in DSD... but that did not prove out.  There are only 2 pure DSD post-production systems today (Sonoma and Pyramix) -- yet despite PCM post-production systems, many studios in fact use DSD for masters...

 

I guess I am agreeing and saying that my high satisfaction with MScaler and Chord applies to PCM. DSD is a small part of my library and I express no opinion on MScaler and DSD.  I can understand finding Chord's broad statements regarding DSD being better rendered through a PCM intermediate state as not being substantiated by independent listeners.

 

All the more reason for wanting to hear impressions vs. MSB, which has invested in both DSD and PCM paths. ?

[Miska]

21 minutes ago, auricgoldfinger said:

Since your software seemingly can do no wrong, when do you plan to address the fact that HQP hangs and must be force closed almost every time the sample rate of a track changes? 

 

Well, this is not HQPlayer thread, I'd like to stick to generic discussion about upsampling filters and such. But you are first to report such. Without knowing your OS/driver and having a log file I cannot help much.

 

23 minutes ago, auricgoldfinger said:

Oh, and there's also the wonderful HQP user interface that causes most users to have to buy a Roon subscription.

 

Is M-Scaler offering a more wonderful user interface? Not that I'd have anything against people buying Roon subscription and using Roon as front-end and HQPlayer only as DSP playback engine. In fact, in HQPlayer Embedded I removed pretty much all user interface and left only simple web interface for some of the configuration things. But added bunch of new features like support for realtime processing from various inputs (analog/digital/upnp).

 

25 minutes ago, auricgoldfinger said:

As someone who has used HQP for 1.5+ years, I'd say it's not exactly ready for prime time.  Maybe it's time to stop taking pot shots at Chord on the M-Scaler thread and focus your valuable time, energy, and experience on your own product.

 

OK, so we cannot discuss any upsampling technicalities except complete isolation from each other?

 

And of course I spend my time and energy on my own product, and have been doing so for past 20 years.

[auricgoldfinger]

2 minutes ago, Miska said:

Well, this is not HQPlayer thread, I'd like to stick to generic discussion about upsampling filters and such. But you are first to report such. Without knowing your OS/driver and having a log file I cannot help much.

 

Search your HQP thread.  I believe it has been reported many times by various users, and you have explained something along the lines of the way the software is designed causes this problem. 

[Miska]

4 minutes ago, auricgoldfinger said:

Search your HQP thread.  I believe it has been reported many times by various users, and you have explained something along the lines of the way the software is designed causes this problem. 

 

If you mean that it stops when you ask it do conflicting things, sure it will do that. It just stops playback and waits you to correct your conflicting request. It doesn't stop if you don't try to make conflicting requests.

 

If you input 44.1 kHz to M-Scaler and ask it to output let's say 1.2 GHz 3-bit data, it will likely refuse to do so. Nothing wrong with that IMO. But I wonder how much flexibility it gives you in first place for selecting output rate and how much it just forces something on you. Maybe you wouldn't face such challenges if I'd reduce configuration flexibility of HQPlayer to the same level?

[auricgoldfinger]

30 minutes ago, Miska said:

 

If you mean that it stops when you ask it do conflicting things, sure it will do that. It just stops playback and waits you to correct your conflicting request. It doesn't stop if you don't try to make conflicting requests. 

 

 

Does HQP hanging (stops playing) and requiring a force close when a 192/24 album comes to an end constitute a conflicting event?  How about when a 96/24 track rolls to a DSD track during a playlist and HQP hangs (stops playing) and requires a force close?

 

30 minutes ago, Miska said:

Maybe you wouldn't face such challenges if I'd reduce configuration flexibility of HQPlayer to the same level?

 

So my expectation that HQP should be able to cycle through a playlist without hanging (stop playing) when music formats change is somehow unreasonable?

[austinpop]

59 minutes ago, Miska said:

 

Well, this is not HQPlayer thread, I'd like to stick to generic discussion about upsampling filters and such. 

 

Why? This is a thread about Chord's M-Scaler, not a generic discussion thread about upsampling  filters.

 

59 minutes ago, Miska said:

OK, so we cannot discuss any upsampling technicalities except complete isolation from each other?

 

You certainly can, but in a product-neutral thread entitled, for example, "Upsampling Technicalities." Feel free to open such a thread.

 

When you do it here, you call into question your motives, since you are in a thread devoted to a competitor's product. Whether you like it or not, you are synonymous with HQPlayer, the product. Do you really want to alienate future customers who may be turned off by what they perceive as you trashing a competitor's product on that product's thread?

 

Why is this so difficult to understand?

 

[Miska]

19 minutes ago, auricgoldfinger said:

Does HQP hanging (stops playing) and requiring a force close when a 192/24 album comes to an end constitute a conflicting event?  How about when a 96/24 track rolls to a DSD track during a playlist and HQP hangs (stops playing) and requires a force close?

 

OK, I haven't seen such and haven't heard of such before this.

 

20 minutes ago, auricgoldfinger said:

So my expectation that HQP should be able to cycle through a playlist without hanging (stop playing) when music formats change is somehow unreasonable?

 

It is for me, and for quite many other people too.

[austinpop]

2 hours ago, Whitigir said:

I used to respect Chord, but losing it every day.

 

Let's say the value of your respect was R on day n, and R-deltaR on day n+1. May I ask what transpired in that day to make your respect drop by deltaR? As far as I can see, no one from Chord, or Rob Watts, has said anything on this thread, so what is changing?

 

Could it be you are allowing yourself to get worked up about this? BTW - if you really want to engage with RW and ask him directly, he seems to be more active in the M-scaler thread on head-fi.

[Miska]

18 minutes ago, austinpop said:

Why? This is a thread about Chord's M-Scaler, not a generic discussion thread about upsampling  filters.

 

OK, I hope then that nobody will compare it to any other product in this thread. It must be discussed without any references or relations to any other product.

 

I join discussion in topics I find interesting.

 

18 minutes ago, austinpop said:

You certainly can, but in a product-neutral thread entitled, for example, "Upsampling Technicalities." Feel free to open such a thread.

 

You know, usually discussion starts somewhere and continues to various directions. I don't see it unusual. Sometimes Chris splits threads to new ones when such happens. 

 

18 minutes ago, austinpop said:

When you do it here, you call into question your motives, since you are in a thread devoted to a competitor's product.

 

I don't see hardware manufacturers as competitors, I don't sell any hardware. I am pretty sure anybody using M-Scaler and playing music from a computer will still have some player software in the computer?

 

18 minutes ago, austinpop said:

Whether you like it or not, you are synonymous with HQPlayer, the product.

 

I am not, I am first and foremost audiophile, who happens to work on HQPlayer among other things. I cannot accept situation where I cannot participate on any other discussion on any forum other than something regarding HQPlayer. Just like actors couldn't discuss anything else than about movies they have acted in. I find such unacceptable.

 

(I hope I'm then also synonymous with all the other projects I've worked on over the past 30 years, that would be incredibly cool!)

 

18 minutes ago, austinpop said:

Do you really want to alienate future customers who may be turned off by what they perceive as you trashing a competitor's product on that product's thread?

 

I don't understand where am I trashing something?

[ecwl]

1 hour ago, Miska said:

I don't understand where am I trashing something?

I think @Miska is very passionate about your philosophical approach to DAC design using software to upsample and noise shape 1fs to DSD512. And there is no doubt HQPlayer is a fantastic product that has benefited many DAC owners (for PCM/DSD type playback/upsampling). But I think sometimes when it comes to something that you disagree with related to Chord/Rob Watts, you can be somewhat dismissive or critical. While it may be true that the Chord/Rob Watts approach is wrong but I think most people give you benefit of the doubt when you talk about HQPlayer, I think the same courtesy can be extended to Rob Watts' comments from Head-Fi forum.

[ecwl]

On 7/24/2018 at 9:54 AM, Miska said:

Chord uses so simplistic modulator that the noise floor looks more like something between DSD128 and DSD256 even though it runs at 100 MHz.

This is based on your measurement of Mojo. It's been pointed out to you that newer Chord products use better noise shapers.

 

On 7/24/2018 at 5:08 PM, Miska said:

Order is just as useless number to describe a modulator as is number of taps to describe a filter.

Higher order modulator makes the increasing noise slope steeper, which in turn puts more demand on the analog filter.

In fact, Rob Watts has said that the Chord DAVE noise shaper does not even fit into the FPGA of the original Chord Hugo. So Rob Watts does recognize your point that more computational power is required.

 

And when pointed out that Pulse Array DAC uses constant switching.

On 7/24/2018 at 5:27 PM, Miska said:

And the DSC1 DAC design I published for playing up to DSD512 (24.576 MHz) uses 32 discrete elements with constant switching...

You believe that it is just the same technology as all other DACs, including yours. When I tried to explain what I suspect Rob Watts meant for constant switching, and the rationale behind switching the same number of elements to reduce switching noise and improve jitter immunity and noise floor modulation, you were dismissive and said

On 7/24/2018 at 6:59 PM, Miska said:

Regardless how many values you want to represent, this is generally nice property of thermometer code (scrambled unary coding) used with SDM; you can represent the same value in many ways.

 

This is not to say you're necessarily wrong. Maybe Rob Watts doesn't have any secret sauce and is just a charlatan talking up his technology. But I think the courtesy might be to assume there may be something he is doing that's different?

[ecwl]

On 8/1/2018 at 4:29 PM, Miska said:

Ahh, and of course use extremely short filters (with only few taps) - completely opposite of Chord's while doing so. And then force feed that to everyone without choice.

 

I'm sure Rob Watts will love that idea.

Sometimes, just because there is a difference in approach to short filter/DSD vs long filter/PCM, there is just no need to make it personal by saying people are force feeding anything. No one would say @Miska is force feeding us DSD. You think upsampling/noise shaping to DSD512 is a better idea, that's cool

[ecwl]

4 hours ago, Miska said:

Last time I checked they had just 120 dB stop-band attenuation, which in my opinion is not enough, or in other words I don't see reason why not do more. That's why I use 192 dB so that it matches at least 32-bit resolution... But anyway, I think there are so many aspects in a filter design, that single parameter like number of taps is pointless without all the other information...

Once again, I am not sure where your 120dB stop-band attenuation comment comes from. Rob Watts did mention that DAVE's WTA filter typically has stop band attenuation at 140dB (worst case 120dB). He said he was surprised that when he turns on the HF filter on DAVE, CD's still sound better. So he recognizes that for reasons he doesn't understand, 140dB stop band attenuation was insufficient. He also said that he may change this in the future. Not sure if it is incorporated into M-Scaler. I think he did because he says he can no longer hear a difference between whether the HF filter is turned on or not when you run M-Scaler into DAVE. That said, we don't know what the stop band attenuation is at now. He did also mention that in Hugo 2/TT 2/Qutest/HMS, when he decimates DSD to 705.4kHz, he uses a stop band attenuation of 220dB.

 

I think ultimately, all of us here recognizes that you, @Miska has a lot of great insights into DAC design and has provided a great service to all of us by creating HQPlayer and has contributed a lot of thoughtful comments to help us optimize our music playback. In fact, I met a couple of new people at our recent local Head-Fi meet and I was going to tell them to try out HQPlayer to pair with their Schiit and ADI DACs. It would sometimes feel nicer to have a positive environment to discuss these fascinating aspects of DAC design.

 

[auricgoldfinger]

@The Computer Audiophile At what point do you declare that a thread is no longer on topic and that posts need to be split off to a separate thread?  Some very respected CA members have politely suggested it would be appropriate for this thread.  A fresh pair of eyes looking at this thread for information on the M-Scaler might be quite confused since there is little informed discussion of the M-Scaler product.  I am curious about the criteria as I am relatively new to CA.

[mav52]

12 minutes ago, auricgoldfinger said:

@The Computer Audiophile At what point do you declare that a thread is no longer on topic and that posts need to be split off to a separate thread?  Some very respected CA members have politely suggested it would be appropriate for this thread.  A fresh pair of eyes looking at this thread for information on the M-Scaler might be quite confused since there is little informed discussion of the M-Scaler product.  I am curious about the criteria as I am relatively new to CA.

 

AS the OP I have on a few attempts tried to keep this thread on topic and not turning into a software thread based on HQP alone.  It drifted but came back on track , kinda , but at the moment its moving yet again far off base.  

 

IF someone wants to create a thread that follows the movement of this thread which appears to be how upsampling works, please do.