Posted: 9:00 p.m. EST, 2/23/98

PARK RIDGE, N.J. -- Sony Electronics has revealed the results of an experimental effort to develop a first-generation digital speaker system. The electronics of the speaker system uses the PCM output of a compact disk (or a Dolby Digital receiver), a bank of DSPs with processing power in the hundreds of Mips, and military-grade D/A converters. The amplified output is pumped into an extremely rigid speaker in a highly damped enclosure. Sony believes this combination of state-of-the-art electronics and transducer technologies will set a standard for audiophile speaker systems.

The goal of the experiment was to define a high-end digital pl ayback system that would use what chief design engineer Dan Anagnos has called "second-generation" digital sources. These include the pulse codes from DVD decoders, set-top boxes and multimedia PCs, as well as the re-sampled codes from CD players. Sony envisions new-generation digital entertainment consoles obtaining PCM-encoded data streams from a variety of sources aside from compact discs. DSPs reformat data from a variety of sources and linearize the response to playback electronics to a new transducer.

The speaker system, constructed in New Jersey, Pennsylvania and Japan under supervision of the Park Ridge, N.J. design team, is said to bridge the gap between the best studio-monitoring and high-end consumer systems. While his team defined the ultimate "cost-no-object" speaker system, said Anagnos, his organization will also examine how manufacturable it would be for widespread distribution. Anagnos estimates that the speaker prototype he developed uses some $60,000 to $70,000 in materials, but tha t a commercial version with 90 percent of the experimental features could probably be built for several thousand dollars.

These features include the ability to execute control algorithms and a variety of filter topologies with a personal computer, in addition to dramatic increases in loudspeaker performance. The performance achievements include a linear-phase response (better than ý9 degree input to output from 15 Hz to 23 kHz), an extraordinarily wide dynamic range (116 dB continuous, 122 dB peak) and a totally flat amplitude response of ý0.6dB from 20 Hz to 20 kHz (-3 dB at 18 Hz and 30 kHz). The on-axis sensitivity is better than 90 dB at 1 meter, with a 2.83 Vrms input

While it relies heavily on DSP technology, the digital speaker eschews 3-D audio, surround-sound algorithms, and other specialized effects. Rather, it attempts to recreate concert-hall realism with just two stereo speakers using conventional audiophile criteria-ultra-low distortion and flat frequency response. Using "cost-no -object" criteria, it selects the best speaker design and materials available for the sound transducer, and uses DSP electronics to ensure that the response of the cones is absolutely linear. The DSPs were specially developed for audio-equalization applications by Sony Electronics in Japan.

Anagnos estimates that the Sony DSPs are 30 times more powerful than the 24-bit Motorola 56000 processor. Six are used in the digital speaker system, to maintain the equivalent 24-bit performance on data sampled at 96 kHz. The primary DSP operation is real-time 16th-order finite impulse response (FIR) filtering on small slices of the audio passband. "We're looking for a 96-dB/octave cutoff-effectively-135-dB stopband attenuation," said Anagnos. "This would require hundreds of thousands of taps if we did this with analog circuitry." The DSP speaker architecture uses separate left data, right data and clock signals to keep jitter and phase shifts at a minimum. The goal is to keep noise, distortion and digital artifac ts below the level of 24-bit A/D and D/A converters.

Speaker manufacturers like Snell Acoustics (Haverhill, Mass.) had used DSP electronics to equalize a high-end speaker system in its home environment. A room with rugs and heavy curtains, for example, would have more of a damping effect on high frequencies than a room with wood floors and plaster walls, and each would effect the perceived sound quality coming from the speaker. The DSP equalization would not be completely adaptive. Rather, an installer (or knowledgeable end-user) would measure the damping characteristics of a particular room, and calibrate the speaker accordingly. The Sony system operates more "open loop," said Anagnos. Initially, a microphone is used to characterize the system in a variety of listening environments and listener positions. But then all the empirical data is "crunched down" in a series algorithms that are used to program the DSPs. Anagnos projects that the user will be able to provide some adaptive correction after the commercial versions of these speakers are shipped, but this would likely be keyboard strokes on a computer used to call forth one algorithm or another (rather than a recharacterization of the speaker in its new environment). The use of software programming, in addition, allows the commercial speakers to be upgraded.

The speakers are intended to accept a variety of PCM digital inputs, from 16 to 24 bits, sampled in any frequency from 44.1 kHz to 96 kHz. A sample rate converter will accept a 16-bit, 44.1-kHz input from compact disks, and reconstitute it as what Anagnos calls a "One-Bit" data stream-an oversampled 24-bit, 96 kHz signal. Separate analog-to-digital converters will allow the speaker to accept wideband analog inputs. In its current incarnation, the digital speaker uses $15,000 D/As from DCS, a U.K.-based radar-systems manufacturer. Though no intent to replace the Sony proprietary DSPs is acknowle dged, Anagnos admitted he was "in conversation" with Burr-Brown (Tucson, Ariz.) about using its audio D/As in a possible commercial rollout.

But it remains uncertain how much Anagnos's experiment will resemble the actual commercial product that Sony releases as a DSP speaker. The transducers (woofers, tweeters and midrange cones) in the new speaker are rigid, highly damped diaphragms. The midrange, for example, is a pure magnesium cone; the woofer is anodized aluminum; the tweeter is a proprietary composition fabric. Sony said these materials provide greater linearity and significantly lower distortion. (It allows, in audiophile terms, a "more-neutral, musical and natural sound quality.") A symmetrical magnetic-field design in the speaker cones provides greater linearity and much higher dynamic range, and reduces distortion at high excursions. The speaker cabinet is designed to reduce acoustic reflections and colorations-a 4-inch thick constrained layer damping wall-not unlike concrete. Thus, in addit ion to their $70,000 cost, users interested in the Sony prototypes must also accommodate the bulk of the speaker cabinets. Each one weighs 572 pounds.

  Go to This Week's News

Free Subscription to EE Times First Name Last Name Company Name Title Email address   Click here for your Free Subscription to EETimes Europe