I have heard many “solutions” to the problems of CD sound, but the only ones that have impressed me are jitter-reduction devices. They do not enhance all discs, and they serve little purpose if you have a top-quality, low- jitter CD transport and D/A converter. But these devices do show that reducing jitter can result in sound that has more natural harmonics, detail, sweetness, and soundstage information.
Most other attempts at CD enhancement demonstrate none of these benefits. There is no clear, consistent correlation between the word lengths and oversampling rates of the D/A converters in CD players and their sound quality; even models built around the same basic chips do not necessarily sound equally good. Most esoteric CD accessories are ex pensive rubbish, producing little, if any, improvement in sound quality. High-priced digital cables usually sound no better than any other digital cables that meet the proper technical specifications. Record companies’ vaunted noise-shaping and other enhancement systems, including HDCD, yield discs that are not consistently better (and sometimes worse) than CDs for which no special claims are made.
Audio Alchemy’s latest anti-jitter system, the DTIPro 32, sells for $1,595, and it is the first device I’ve heard in a long time that addresses the basic problems of musical real ism in CD sound. Let me briefly summarize its operation: A phase- locked loop (PLL) locks onto the in coming bitstream. A voltage-con trolled crystal oscillator (whose center frequency has been carefully chosen to avoid interference within the audio band) then locks onto the PLL’s output signal, to reduce jitter; the oscillator also provides a master clock output that is mixed into the signal that feeds the D/A converter. The data signal is processed by a 32-bit Texas Instruments DSP chip, using an Audio Alchemy resolution-enhancement interpolation algorithm. This algorithm is stored in an EPROM memory chip that can be replaced as improved algorithms are developed.
Physically, the DTI Pro 32 is a small black box. On its rear panel are three signal inputs: a BNC coaxial connector that can easily be converted to an RCA input, a Toslink input, and an I2S bus input for connection to Audio Alchemy CD transports. (The 1 bus, used within many companies’ digital audio components, is also used by Audio Alchemy for digital links between components.) You can substitute an AT&T input for the Toslink at an additional cost of $179. The outputs are BNC, AES/EBU, AT&T optical, and I2S. The external power supply connects to another rear-panel jack.
Inputs are selected by a front-pan el pushbutton; three LEDs show which input has been chosen. The other panel button (“Phase”) selects normal or inverted polarity and also has LED indicators. Between the controls are two more sets of LEDs; one set shows when the unit is receiving, transmitting, and processing signals, while the other set indicates when the PLL and crystal oscillator are properly locked to the signal. Pressing the buttons in combination lets you change the dither setting to select word lengths of 16, 18, 20, 22, or 24 bits for output to your D/A converter; you can also turn dither off. (Anytime a digital signal is requantized, as in Audio Alchemy’s resolution-enhancement process, it should be redithered to minimize distortion.) Since the DTIPro 32’s resolution-enhancement algorithm conflicts with HDCD, another switch combination turns the enhancement off for HDCD reproduction; this must be done each time you use HDCD.
The DTI Pro 32 is not complicated, but not all of its features are obvious. You must follow the instructions in the owner’s manual to set it up, although basic setup (including reading the instructions from cover to cover) takes all of five minutes.
Deciding on the proper operating mode requires extensive listening. Start with the word length set to 18 bits, then try 20 bits, and so on. Many audiophiles who have 20-bit D/A converters instinctively go for the higher settings, but the 18-bit setting may sound better. (Audio Alchemy offers advice by phone if you’re unsure about the optimum setting for your system.)
The improvement the DTIPro 32 makes depends significantly on the rest of your CD playback system and on the quality of individual discs. For example, I found the benefits more audible with old CD transports than with top-quality newer models.
By contrast, the sonic improvements were more musically significant with really high-quality DIA converters, such as the Mark Levinson No. 30.5 and Theta Digital DS Pro Generation V. But I heard on minor differences when I used the DTIPro 32 with medium-quality 1-bit D/A converters. It sometimes made a striking improvement with medium-quality multibit converters, but there were also cases where I couldn’t properly hear the DTI•Pro 32’s benefits.
The improvement made by the DTIPro 32 also varied from CD to CD. The benefits seemed to be governed by the sonic quality or clarity of the recording and not by the disc’s age or whether it was made from a 20-bit master or noise-shaped. The DTIPro 32 does not solve all of CD’s sonic problems. It cannot alter the word lengths and sampling rates of recordings or make CDs sound like the best digital tapes. However, when I played recordings of solo violin or grand piano through it, I noticed an impressive improvement in their harmonic integrity. The common high-end terms for such improvements, “sweetness” and “air,” don’t really apply. The DTIPro 32 did not sweeten or add “air”: It provided added definition that made properly recorded instruments sound more real.
With other instruments, the sonic benefits varied. Although solo guitar often sounds very good on CD (better than violin, for some reason), the DTIPro 32 still made a difference with a number of guitar recording. For example, Julian Bream’s recordings, which often sound less harmonically realistic on CD than on LP, gained a bit more life when I used the DTIPro 32. And although I rarely find the harpsichord musically convincing on CD, LP, or tape, it did sound more real on CD with the DTI-Pro 32.
As an ex-drummer, I’ve noticed that CD often has considerable trouble reproducing brush, cymbals, triangles, and drumhead textures. The DTIPro 32 gave me a lot less sonic improvement with these instruments than with the others I have just described, but it did make a difference, particularly with really clean acoustic jazz recordings. Again, there seemed to be a significant improvement in musical resolution and real ism. Bass dynamics also seemed to improve, though this may have been the result of improved transient and soundstage-detail resolution rather than greater or more extended bass energy.
The Audio Alchemy processor also improved the resolution of orchestral detail on some discs; many Reference Recordings orchestral CDs sounded better and more natural with the DTIPro 32’s processing than with HDCD decoding.
With voice, the Audio Alchemy processor made choral textures and definition a bit clearer, and it often improved the fine de tails of soprano and tenor voices. (The effect with baritones was less clear.) I did not, however, find the improvement as striking with voice as with instruments, perhaps because my ear is tuned more to classical instrumental sounds.
My sons preferred listening to many of their electric rock and jazz CDs through the DTI-Pro 32. I’ll rely on their opinion, since I don’t particularly care how unnatural the unnatural gets.
The DTI-Pro 32 is one device that really makes a difference in CD sound. The aging standards on which CD is based set a ceiling on its sound quality. This becomes more and more apparent as superior equipment comes into common use. Audio Alchemy’s DTI-Pro 32 can’t break though that ceiling, but it can make sizable cracks in it. If your main concern is musically natural sound, the difference it makes can be vital.
Also see: This Stereophile's article