oversampling to DSD with SHARC?

Hi all,

as owner of a sony sacd player (xb940) and numerous listening tests i came to the conclusion, that redbook cd can be shifted to a higher level via upsampling to dsd. My player is modded with Tent Clock, PS - mods and a Zapfilter Mk2.

I am NO digital expert at all, but a musik lover who wants the best bang for the buck. I cannot afford Barclay or Burmester transports but I think a cheapskate like me can get comparable or even better quality by ripping the cd to harddisk and play it via foobar and an external usb- device. And here the problems arise, IMO

1.) All usb devices are working in synchronous mode so the phase difference between the clock beneath the soundcards PCI controller and the master clock (at the dac) have to be dealt with. Only Spact by BB seems to solve the problem, but its restricted to 48Khz sampling rate. (BB-Ti PCM 2706 with I2S output)

2.) The I2S protocol seems to be the best digital interface, but its normally only used for datatransfer from receiver to dac. Two known exeptions: Elso Kwak's "I2S - direct" and the now defunct audio alchemy DTI 32 jitterkiller with proprietary I2S output


3.) There seems to be no literature/whitepaper how sony, philips, meitner or dcs do the upsampling from redbook to DSD.

It's my firm belief (from what I'm hearing) that upsampling to DSD is the key to much better ambience and treble, and helps redbook a lot - avoiding "listening fatigue"

What I want to realize: To send the 44,1 or 96 kHz data from the PC via usb - as bitperfect as possible - to the asynchronous USB receiver, then asynchronous resampling by means of a suited DSP should take place to transcode data to 24/176,4(192) or even better DSD.

The data should then be digitally filtered via software filters (un-like in the Sony VC24 chip) and then be sent as DSD signal (or I2S) to the DAC and outputstage.

In short words: I want to replicate the Sony /Philips/Meitner/DCS upsampling with commonly available chips/DSP's

Anyone out there who can help me a bit?

Cheers, Helmut
 
It's certainly possible. The DSD process isn't that hard to do, and a newer SHARC should provide plenty of horsepower for the interpolation part (i'm not sure what's involved in generating 'DSD' dither)

But I've got a couple of issues with this:

(1) Almost every modern CD player out there has a sigma-delta DAC inside, which does almost exactly the same thing as DSD conversion.

In fact, earlier sigma-delta DAC chips would interpolate 44.1KHz audio up to a much higher rate and chop it off to only one bit, but modern DACs use multibit architectures which improve quality and avoid many of the problems associated with single bit sigma delta. And DSD is single-bit...

So you could get a CD player with an earlier sigma-delta chip and just use that instead. ;)

(2) The advantage of DSD is that extra information is carried on the disc - the 22.05KHz nyquist limitation isn't there, and 20-22KHz filtering isn't required on either the studio or playback ends. "DSD-ing" the output of a CD will give you the same content that is on the CD, which has these limitations included.