Technical Description - 4096 WTA Filter

The WTA filter algorithm has taken twenty years of research to develop. It solves the question as to why higher sampling rates sound better. It is well known that 96 kHz (DVD Audio) recordings sound better than 44.1 kHz (CD) recordings. Most people believe that this is due to the presence of ultrasonic information being audible even though the best human hearing is limited to 20kHz. What is not well known is that 768 kHz recordings sound better than 384 kHz and that the sound quality limit for sampling lies in the MHz region. 768 kHz recordings cannot sound better because of information above 200 kHz being important - simply because musical instruments, microphones, amplifiers and loudspeakers do not work at these frequencies nor can we hear them. So if it is not the extra bandwidth that is important, why do higher sampling rates sound better?

The answer is not being able to hear inaudible supersonic information, but the ability to hear the timing of transients more clearly. It has long been known that the human ear and brain can detect differences in the phase of sound between the ears to the order of microseconds. This timing difference between the ears is used for localizing high frequency sound. Since transients can be detected down to microseconds, the recording system needs to be able to resolve timing of one microsecond. A sampling rate of 1 MHz is needed to achieve this!

However, 44.1 kHz sampling can be capable of accurately resolving transients by the use of digital filtering. Digital filtering can go some way towards improving resolution without the need for higher sampling rates. However in order to do this the filters need to have infinite long tap lengths. Currently all reconstruction filters have relatively short tap lengths - the largest commercial device is only about 256 taps. It is due to this short tap length and the filter algorithm employed that generates the transient timing errors. These errors turned out to be very audible. Going from 256 taps to 1024 taps gave a massive improvement in sound quality - much smoother, more focused sound quality, with an incredibly deep and precise sound stage.

The initial experiments used variations on existing filter algorithms. Going from 1024 taps to 2048 taps gave a very big improvement in sound quality, and it was implying that almost infinite tap length filters were needed for the ultimate sound quality. At this stage, a new type of algorithm was developed - the WTA filter. This was designed to minimize transient timing errors from the outset, thereby reducing the need for extremely long tap lengths. The WTA algorithm was a success - a 256 tap WTA filter sounded better than all other conventional filters, even with 1024 taps. WTA filters still benefit from long tap lengths; there is a large difference going from 256 taps to 1024 taps.

Currently the DAC64 uses 1024 taps. The filters are implemented in FPGAs (Field Programmable Gate Arrays) using a specially designed 64-bit DSP (Digital Signal Processing) core. The Chord CD transport will use a 4096 taps WTA filter. This will further improve the ability of the filter to reconstruct transient timing. We know this to be true from the DAC64, which already increased the tap lengths used from 516 to 1024. What this will mean in terms of sonic performance is when combined with the DAC64 we can achieve perceived improvements in bass, rhythm, timing and sound staging.

All of the above innovations are implemented in Xilinx Spartan series FPGA's. These FPGA's can offer in the CD transport 400,000 gates per device, and merely updating the EPROM memory chip can easily change the design, thus future proofing is assured.

Sampling Frequencies

As already explained above it is well known that the higher the sampling frequency, the better the soundstage quality i.e. 96kHz recordings (DVD-A) sound better than 44.1 kHz (CD) recordings. However we at Chord are using up-sampling correctly, the CD Transport will up-sample the signal before transferring it into the Digital to Analogue converter by whole factors, not derived clock signals i.e. 96kHz or 192kHz. We are outputting from the transport, true multiples of the original sound recording frequencies:

44.1kHz x 2 = 88.2kHz
88.2kHz x 2 = 176.4kHz

We have specifically chosen these frequencies because they do not risk adding any digital noise into the signal that 96kHz and 192kHz can suffer from because they are not truly up-sampled by whole factors.