and Rob Watts himself talks about the Quartet for those who are interested..https://www.head-fi.org/threads/chord-electronics-quartet-m-scaler.980480/
"
I am going to start this thread with a very bold claim - which I can justify technically, and I hope will be confirmed by future listeners to the Quartet M scaler and the pulse array ADC.
Quartet represents the most radical improvement in digital audio technology this century.
There we go, I have got that off my chest.
And I believe it does this for two reasons - the ground breaking pulse array ADC built into Quartet, and the new Blackbird WTA interpolation filter, which reconstructs transient timing to unprecedented accuracy. The interpolation and ADC represents two sides of the same coin, which is the fundamental flaw of conventional digital audio. The flaw limiting performance is the conversion from a continuous analogue waveform (the output of a microphone for example) to sampled data which is no longer a continuous signal, but snapshots at regular intervals, which is the work of the ADC; then with the DAC we have to convert from sampled data back to a continuous waveform to drive headphones or a loudspeaker.
But this conversion from continuous waveform to sampled data then back to a continuous waveform has a huge problem - conventional digital audio destroys the timing of transients. It's like putting a steak through a mincer, then reconstructing a steak back from minced beef, and expecting it to be the same.
The timing of transients is the most important aspect of audio engineering, and yet it's ignored by the audio technical community. It's vital because what we perceive is not the output from our ears, but an illusion created by the listeners' brain; everything we perceive has transient timing at it's heart. We perceive pitch, instruments as separate entities from transient timing, timbre, placement of instruments in space, and of course, the starting and stopping of notes from transients.
So what is the problem with DACs? It's the fact that the transient timing is the waveform in-between individual samples from the ADC; what the DAC has to do is recreate the original timing of transients by using the interpolation filter (all DACs have digital interpolation filters even NOS DACs) - and conventional DACs have poor reconstruction accuracy. Transients are constantly shifting in time being early or late, which confuses the brain, destroying perceptions of timbre, pitch, and instruments as separate entities located in space.
Sampling theory tells us that we can perfectly re-create steak from the minced beef analogy if we use a sinc function filter; but a sinc function filter is impossible to implement as it requires an infinite amount of processing. Real conventional interpolation filters have extremely limited processing, and are nothing like sinc functions, so create transient timing errors, making transients constantly shifting early or late, destroying the perception of music.
To improve sound quality, I developed the WTA interpolation filter. This reached it's zenith in performance with the Hugo M scaler, which had advanced processing with 1 million taps. The WTA filter in the M scaler was identical to the ideal sinc function interpolation to better than 16 bits; this guaranteed transient timing reconstruction to better than 16 bit accuracy. But in developing the WTA filter, it used a unique formula with five variables; these variables were adjusted to 10 parts per million accuracy by thousands of listening tests, with each variable having a sweet spot. In 2018, fundamental research started - a multi million USD project that I invested in - which led to Quartet some 8 years later, and this research was trying to understand why the WTA variables were so finely balanced. It turned out the DAC interpolation filter has two ways of damaging transient timing, and they were in opposition to one another, hence the fine balancing of the WTA variables. Also, with the existing WTA, we would need impossibly long interpolation filters for adequate reconstruction. Having understood the issues behind it, I invented a new filter structure - the Blackbird WTA. This uses multiple stages of parallel filtering, but crucially allows treating the two ways of damaging transient timing separately; this now allows ideal reconstruction, only being limited by processing power and internal accuracy. The only downside to Blackbird WTA is it needs more FPGA processing power, due to the multiple stages of parallel filtering.
The technical benefits of the Blackbird WTA in Quartet are:
1. Blackbird WTA provides at least a ten fold improvement in transient timing accuracy over the original WTA for the same tap length, and some transient timing errors are completely eliminated
2. Quartet has 4M Blackbird WTA taps, with a latency (delay) of 3 seconds. This latency can be adjusted for use with video
3. All filtering is direct implementation, not through transient damaging FFT convolution unlike software up-samplers
4. Coefficients are optimised and calculated to 64 digits accuracy, then aggressively noise shaped to preserve accuracy
5. Four filter stages are employed with unparalleled resolution
6. Advanced HF filters to completely eliminate HF distortion and noise from the recording, which improves sound quality
7. Quartet has five 200T FPGAs with a total DSP core capacity of 3,700 and over 2 million lines of code
Quartet also has subjectively loss-less EQ, with a ten band octave shelf control. It features 4 stage isolation on the 705/768 kHz outputs, with an advanced separate PSU that provides battery level isolation from the mains by using a new type of RF filter.
The other side of the coin with the continuous signal to sampled data back to continuous signal problem, is the ADC. This is also subject to transient timing errors; in this case it's down to errors due to aliasing. Aliasing is a nasty form of distortion when out of band HF signals from the ADC noise shaper is distorted into lower frequencies; the presence of aliasing damages transient timing. All conventional ADCs use half band filters, which create aliasing errors at a peak of -6dB from 0dB HF content. With Quartet pulse array ADC, advanced decimation filters ensure aliasing has been eliminated from the 104MHz pulse array noise shaper output down to 768kHz or 192kHz. Moreover, the pulse array ADC has all the benefits of pulse array DACs - zero measurable noise floor modulation and extraordinary small signal accuracy. Sound quality tests using non Dolby master tape as a source, and comparing the Quartet ADC against a high end pro ADC, has revealed huge sound quality improvements, even audible through a mobile phone loudspeaker!
Quartet's sound quality benefits from digital sources when played through Chord DACs is similarly not small. Instruments sound real and tangible, with a much more life-like sense of timbre. Reverberation and the sense of space is dramatically improved. Perception of pitch, particularly with bass, is much better. The music also has a much more natural sense of flow - tempo without Quartet sounds mechanical and disjointed. All of these sound quality effects are consistent with the understanding of the importance of transients from psychoacoustics and the reduction in transient timing errors that Quartet has.
Quartet M scaler is being launched later this week at the High End Vienna (June 4 to 7, 2026). UK retail is £25,000 and units will be available shortly.
I will be doing two seminars at London CanJam and Socal CanJam about the Blackbird WTA and the pulse array ADC, and I will publish the seminars on this thread after Socal."
signals.uk.com
