Interview with Chord’s Rob Watts

Stumbled onto this and found it enlightening. It’s a 4 part series.


Well, at least he’s honest about using his products the most and not having enough experience with R2R.

Let’s say I’m all about measurements. I don’t get why I should waste my money on Chord products when a PCM to DSD converter + Topping E30 or D90 set on Direct DSD gives me the best quality I could get.
Something like HQPlayer has lots of different filters and algorithms. Even Neutron Music Player has better filters than most of these DACs. I can get a customized sharp filter with lots of attenuation with Neutron. Even D90 can’t do that.
D90 has a spike in ultrasonic frequencies which is the result of poor attenuation.

R2R is more about personal taste. I don’t like R2R but I can see with certain genres it sounds subjectively better.

Chord DACs have two main important pieces, the filter that tries to reconstruct the original sample, and the Delta-Sigma conversion. In those videos he explains why his approach on the delta sigma conversion is superior to DSD, and it has to do mostly with transients. He focuses a lot on the signal reconstruction, timing and transients.

Chip DACs like ESS and AKM have simpler filters, and everything is packed in a single chip.

I believe Chord DACs are overpriced but they might be the best Delta Sigma DACs you can get.

That’s what HQPlayer has too. The main differences between Chord dacs and DSD is 1) Chord is multibit vs one bit DSD 2) Chord has to have Dynamic Element Matching

Multi bit brings back the main problem of R2R dacs that he doesn’t like: non-linearity. So he has to use DEM to randomize the noise.

DSD512 has so much bandwidth that the DSD noise would not be a problem. Also, DSD is inherently linear.

Plus, Chord dacs’ problems are more on the analog side than digital. They don’t measure as good as Topping dacs.

About transition, people who hear softness or weird things with DSD don’t apply a good low pass filter on the PCM side. Even some people who release DSD masters forget to do that when converting from DXD to DSD.

The proper way to do PCM/DXD to DSD is

  1. oversampling to at least 176.4khz
  2. digital lowpass filter
  3. delta sigma modulation

Because there’s no signal in ultrasonic range of the PCM file, people think there’s no need for lowpass filter.
They forget about imaging that happens in ultrasonic range. Anti-imaging or reconstruction filter is basically oversampling+lowpass filter.

All dac manufactures even in R2R era knew about imaging.

I enjoy Chord products very much, so much so that i put out my hard earned funds to own one of his pieces. I also have ears that work And MY OWN personal sound preferences along with a varied sampling of equipment to pair different Products with each other to find synergy OR reproduce different sound signatures to match my listening environment, mood and particular flavor choice for the moment.
Engineers can measure all they want and debate the merits of one technology over another, consumers can freely pick amongst the varied approaches to music reproduction, musicians can take their artistic talents and vary their own approaches with equipment, choices of venue and Ultimate sound quality they choose to capture for posterity or profit. What makes this hobby so addictive to me is that i can choose to listen to the music, the equipment, or combination of both depending on my own wants, desires, abilities and ultimately lifestyle choices.

I mentioned that i have ears which work, and this is an important part to all this because my ears are tuned and in sync with my character, mood, and listening abilities. Some of us have much less ability to hear, discern and appreciate the fine nuances found in musical reproduction. All of us have varied tastes based on culture, environment and simple preferences. This is why there is so much varied equipment with capabilities out there. Yes absolutely many pieces of equipment, listening environments, and methods of reproducing music really really work flat out better than others But ultimately the fine differences are in the details many times minute as they may be.

Have at it ladies and gentlemen, discuss the merits and negatives of each piece in detail as long as you enjoy yourselves the net result is a benefit to all. This was a good interview, thanks for posting @A_Jedi :+1:

Did you watch all 4 videos? He goes into some detail about measurements and human sound perception. The best bits about that are in part 4.

Is Chord the equivalent to Apple of the audio world? Objects of beauty, good at what they do, but disproportionately priced for what you’re getting?

Would love to give a listen to a higher end Chord DAC some day. The longer I’m in the hobby the more I realized that I’m not main stream with the signature and styles of sounds I prefer.

I’ve watched the fourth part. I’m not saying he’s wrong and Chord is bad. I’m saying much cheaper products (hardwares and softwares) offer similar to or even better than what Chord offer.
He talks about things we don’t know about the brain and how similar measurements sound different (which I agree. THX and Topping A90 is one example), but jitter, noise floor, voltage, etc. are measurable.

His dacs have some Mhz sample rate. DSD512 is 22.5792Mhz. That’s much more than 3Mhz of dac-chips. I agree with his statement about transition. But HQPlayer+Topping D90 is much cheaper than his products. D90 is one of the best-measured dacs available, if not the best.

Here’s the jitter of Chord Mojo with optical:


Here is of Topping D90 with both USB and coax:


That’s like, 20db difference? He can say some unmeasureable things influence the sound, but this dirty optical is clearly measurable.

@15:29 he says that PC isn’t powerful enough for oversampling process. Come on dude! It’s not like HQPlayer is taking 100% of CPU usage. Is he saying Chord’s FPGA is faster than the newest Intel Core i9? What?
It’s like Steve Jobs and defending iPhone’s 3.5 inches display. Chord is indeed Apple of audio gears.
Some companies ignore measurements and focus on musicality. Chord isn’t one of them. When he brags about his special algorithms, these dirty measurements shouldn’t be there.

He said PC isn’t powerful enough unless it’s using a graphics card for the processing. Graphics cards are actually MUCH more capable of complex processing than any Intel cpu. It’s why they’re used for crypto mining.

Just to be clear - I’m not advocating that he’s the arbiter of what is true and what isn’t. I simply found many of the things he says interesting.

Intel CPUs eat the most powerful Chord with ease. Mining is a different story. The whole audio processing is rather simple. If he was talking about Chord’s hardware like special capacitors or resistors, I wouldn’t get annoyed. But software and FPGA?

I’m not fighting lol. I don’t mind companies making a brand by saying weird stuff. That’s what Apple does. I actually agree with most of what he said except minimum phase being bad.

Chord does not use R-2R. Rather it’s an FPGA that implements its own delta sigma. Just an FYI.

When did I say Chord has anything to do with R2R?

Honestly, I’m reading my sentence over and over and don’t get how you thought I didn’t know Chord isn’t R2R.

Rob is honest about using his products (which means Chord lol) and he doesn’t have enough experience with using R2R. I said that because the host prefered some R2Rs to DS chips although R2R measure worse. In the end, Rob was like whatever, I listen to my products becuase I believe they’re the best.

A couple of things I found particularly interesting:

Our brain can process sound at a speed of at least 250KHz :scream:

RFI entering the system is bad not because we can hear it but because those frequencies are difficult for amps to reproduce so overall performance suffers. Time to cover everything in tin foil lol.