Meridian Unplugged - Powered by MrTechGuy.co.uk Forums Discussion Meridian - Systems Anyone heard apodized digital through DSP6000s?

David,
http://www.stereophile.com/cdplayers/meridian_8082808i2_signature_reference_cd_playerpreamplifier/

I will provide some clippings that help me understand it.

along the way, the data are subjected to a new kind of reconstruction filter that Meridian calls an "apodizing" device.

"Apodizing"
A pair of technical papers in the March 2004 issue of the Journal of the Audio Engineering Society (Vol.52 No.3) indicated the direction Meridian would be taking with the 808.2 (footnote 1) In "Coding for High-Resolution Audio Systems," Bob Stuart examined what was required to get the highest audio quality from a digital system, while in "Antialias Filters and System Transient Response at High Sample Rates," Peter Craven examined the nature of the ubiquitous anti-aliasing filter (in A/D conversion) and reconstruction filter (in D/A conversion) (footnote 2).

Keith Howard had explored this second subject for Stereophile in January 2006. While the linear-phase Finite Impulse Response filters routinely used in digital audio can offer superb performance in the frequency domain, in the time domain they introduce preresponses that occur prior to the main peak of the impulse response, something that never occurs in nature and thus will not be masked by the music. The brevity of these preresponses that occurs with high sample rates has led some commentators to decide that this is why high sample rates produce better sound quality, and not the octave extension in bandwidth itself. On the face of it, however, this improvement seems paradoxical: the preresponse comprises ringing at half the sample rate (called the Nyquist Frequency), which, even with CD, is above the limit of human hearing. But the preresponse does appear to have an audible effect, perhaps because the human ear/brain system acts as a wavefront detector rather than as a spectrum analyzer.

What Peter Craven explored in his paper was the idea that you could process the audio data with a minimum-phase low-pass filter whose first null is at the original Nyquist frequency. This will therefore kill all the ringing from the original A/D converter, replacing the preresponse ringing with postresponse ringing from the new filter. As this new ringing would a) be natural in that it occurs after the event, and b) takes place entirely within the human auditory system's masking interval, the data would be audibly cleaned up. The tradeoff, however, is that the new minimum-phase "apodizing," or windowing, filter has to begin rolling off below the original Nyquist frequency, thus sacrificing a small fraction of the top octave of the original passband in order to optimize the time-domain performance.

The apodizing filter used in the 808.2 and 808i.2 was optimized using listening tests. It is realized using a DSP chip operating at 150MIPs with 48-bit precision that also performs the upsampling. It appears that the CD data are first upsampled to 88.2kHz, then processed by the apodizing filter which has a null at 22.05kHz. This is the datastream that is fed to the Meridian's SpeakerLink digital output. The DSP chip then further upsamples the data to 176.4kHz to feed the player's DAC chips.

Were these benefits the result of the 808i.2's apodizing reconstruction filter eliminating the time smear from a recording's original A/D converter? I'm tempted to say yes, except that there are many other aspects of digital music playback that also have to be gotten right. But now that I've discovered anew the sonic beauty of so many CDs in my collection, it's going to be difficult to relinquish the Meridian when the loan period is over.

My final comparison was intended to provide an answer to this question: How does the CD playback of the Meridian 808i.2, with its apodizing filter, stand up against genuine high-resolution recordings played through the Ayre? I burned the 24-bit/88.2kHz two-channel masters of two of my recent recordings—Attention Screen's Live at Merkin Hall (Stereophile STPH018-2) and Cantus's While You Are Alive (Cantus CTS-1208)—onto DVD-Audio using Minnetonka Software's Discwelder Bronze program, then played them back on the Ayre. The CD versions were played on the Meridian with as close to perfect time synchronization as I could manage. Again, levels were matched to within 0.1dB at 1kHz.

Having lived with these recordings for months and played the files through conventional D/A converters such as the Benchmark DAC1 and the Logitech Transporter, I'm convinced that, as good as the CDs sound, the hi-rez masters go that essential step further in soundstage solidity and image palpability. So it was with some surprise that I found it very difficult—even impossible, at first—to hear any difference between the two players: "Red Book" audio through the Meridian didn't sound appreciably different from hi-rez through the Ayre. Tonally, spatially, the presentations were virtually identical. In the end, there was a difference, though not one I was expecting: The hi-rez playback had slightly better defined and slightly more extended low- and mid-bass with the bass guitar, tom toms, and kick drum on Live at Merkin Hall, which in turn led to a very slightly less congested lower midrange.

Ultimately, my preference was still for the hi-rez versions of the music, but it was a closer-run thing than I had expected.

To examine the behavior of the Meridian's "apodizing" reconstruction filter, I created a 44.1kHz-sampled test signal comprising a pair of positive single-sample pulses spaced approximately 5ms apart. The first pulse of the pair triggered my digital oscilloscope, which I'd set to "one-shot" mode so that the player's reaction to the second pulse appeared in the center of the 'scope's display. (The 'scope lacks an anti-aliasing filter; provided sufficiently fast sampling is used, it accurately displays time-domain data without obscuring those data with its own impulse response.)

Fig.1 shows the impulse response of a conventional finite-impulse-response low-pass filter, that used by the Benchmark DAC1 D/A processor. You can see that there are symmetrical patterns of ringing before and after the peak of the impulse. (I call it "ringing," but, as I'm sure engineer readers will point out, the impulse shape actually maps the values of the digital filter coefficients.) By contrast, fig.2 shows the 808i.2's impulse response. There is no ringing before the main peak, but then about 1.5ms of damped ringing after it. While the absence of pre-ringing very likely contributed to the 808i.2's fine sound, the post-ringing was both higher in amplitude and lasted a little longer than I was expecting.

Fig.1 Benchmark DAC1, response to single sample at 0dBFS, 44.1kHz sampled data (4ms time window).


Fig.2 Meridian 808i.2, response to single sample at 0dBFS, 44.1kHz sampled data (4ms time window)