DSP

Digital Signal Processing (DSP) has come a long way since its beginnings in the 1980s as a reverb-manipulating technology for studios. More and more “audiophile” manufacturers—Cambridge Audio, Cabasse, Devialet, Dutch & Dutch, KEF, Naim, Klipsch, Totem, PSB, McIntosh, to name a few—have embraced DSP technology in their products as a means for the consumer to improve their listening room’s acoustics so they can get the most out of their equipment. By some accounts, this may be just the beginning.

Digital Signal Processing in Audio

DSP can be applied to audio during studio production and home playback. DSP chips, which are commercially available, are specialized microprocessors (computers on a chip) optimized for digital audio that can be programmed to perform various forms of signal manipulation. 

Here’s a partial list of what DSP can do:

• Operate like a filter or equalizer in the frequency range to alter the response in several ways.
• Function as an active crossover by means of digital filters that can also provide proper timing to the individual drivers in a loudspeaker. 
• Equalize audio output based on the sound recorded from the listening space to correct for room acoustical issues.
• Be used to change the timing of certain frequencies by speeding them up or delaying them. These effects can improve imaging and soundstage integrity, depth, and width. 
• Provide not just room correction, but listener correction. For example, DSP can correct for where you’re seated.
• Make sounds at low volume more intelligible, using a bespoke Fletcher-Munson curve.
• Adaptive noise cancellation, now quite common in headphones.

DSP use is already well-established in studios, but its use in playback has been gaining momentum. For example, music management software Roon has incorporated a DSP function called Muse, which can be accessed via the Roon app on a computer or smartphone. It can manage audio headroom to prevent clipping, add crossfeed (for headphone listening), convert sample rates (e.g. upsampling), and it offers a parametric equalizer (see screenshot 1) that allows for customizing sound to the listening room or certain audio tracks. More advanced users may also want to dabble with Muse’s procedural EQ and room setup features.

Industry Insights

To better understand what DSP can do and what the future of DSP might hold, I spoke with several manufacturers who are using DSP in their own products, or aspire to do so one day. I also tried to find a few cynics who might tell me why DSP is a bad idea, but to no avail. Does DSP have a future in the high-end? Or is it mostly a niche industry reserved for the pro-audio or budget consumer markets? Let’s see.

An Audio Legacy

Legacy Audio’s Bill Dudleston has been designing and building loudspeakers since the early ’80s. He was an early proponent of DSP processors for speakers, and his Illinois-based company offers two such products for consumer use, the Wavelaunch and Wavelet. The units, which can be used with speakers other than those by Legacy Audio, are digital crossovers that can function as parametric equalizers to remove what Bill calls a tsunami of bloated bass below 500 Hz that tends to accumulate in rooms due to reflections off walls, floor, and ceilings.

Calibration with a Wavelaunch or Wavelet is done using a microphone placed 4 feet directly in front of a loudspeaker while a sound generator sweeps from 10 Hz up to 20 kHz. The microphone picks up direct and reflected sound, which is converted to a data set and then sent to the Cloud where mathematical calculations are done to provide the algorithms used by the processor to correct the room response, first by fixing time domain issues, then by altering the frequency domain. Bill says the end result is a better stereo image, greater soundstage depth, and detail that was previously inaudible due to low-frequency bloat. The DSP algorithms work to allow accurate reproduction of the sound waves that reach the listening position, but also throughout the room.

Bill compares the potential for improvement using DSP to visual images we see on television. He asks, “Who would want to go back to the era of picture tube television, which was state-of-the-art in the 1980s? Today’s TVs are so good, they’re comparable to the best theatre screens today. And video contains much more (data) than audio.

“The problems in audio are different than in video, but there are solutions,” he continued. “Time alignment is one area that speaker manufacturers have struggled with for years. Yet, DSP makes it easier and better than simply moving the tweeter back from the front plane of the speaker baffle. Let the tweeter be in front, but delay tweeter timing so it aligns with the woofer. How much easier and better to do that than with DSP?”

Bill provided some before-and-after images (screenshot 2) to show how his processors help remove unwanted resonances.

He explained: “The wavelet plot shows the cleanup of the multipath reflections below 500 Hz that discolor playback [bottom image]. The images compare the frequencies heard in an uncorrected room to one with DSP room correction. Here, we see that the corrected room contains much less energy at the lower frequencies than what we see in the uncorrected room. This (uncorrected sound) has the effect of covering up much of the detailed information present in the audio signal, by making the sound more bloated with low frequencies that were not present in the original recording, nor due to the design of the speaker.”

Screenshot 2 shows the frequency range (from 20Hz to 20 kHz) on the horizontal axis. The vertical axis is a time scale in milliseconds (zero at top). The colours represent intensity of the sound field, the deepest red being the strongest. Notice the large tongue of colour shown on the first graphic (uncorrected), which represents a strong and delayed wave of reflected sound in the frequency range of roughly 90 to 200 Hz. The sound is clearly in the hearing range of even someone with poor hearing. The corrected audio shows none of that bloated overhang we see in the uncorrected sound. That is the strength of DSP room correction.

Where does Bill see DSP headed as far as the audio hobby is concerned? Most significantly, in standmounts. Just as DSP has become ubiquitous in studio monitors, Bill believes that consumer standmounts equipped with internal power amps and DSP crossovers will become standard. Passive crossovers will cease to exist. This will ultimately make affordably priced audio gear much more palatable to the discerning listener.

The Danville Approach

Minnesota-based Danville Signal offers DSP solutions that can be implemented by speaker manufacturers or individuals. One of the company’s products is a preamp / 8-channel DSP crossover called the dspNexus 2/8, which, as its name implies, can convert 2 stereo channels into 8 channels. It uses upgradable internal modules, and can be used on a combination of individual speaker drivers or external subwoofers. Danville has already implemented its DSP designs in monitors used by studios. To install the dspNexus 2/8, one must disconnect the speaker’s internal crossover and use individual amplifiers to drive each of the speaker’s drivers. That can be achieved rather easily in some speakers, such as the Magnepan MG 1.7i speakers, as demonstrated by Danville at this year’s AXPONA.

To illustrate the effectiveness of its DSP approach, Danville used a passive-to-active switch on the Magnepans’ crossovers. Since each Maggie employs three planar drivers, one amp was used for full-range passive mode and three amps per channel were used for DSP mode. Identical amps were used throughout. According to the presenter—company owner and CEO, Al Clark—no room calibration was performed, but the speakers were equalized for a flat response. Pink noise was played for frequency adjustment and a few bass traps and absorbers were placed throughout the room. Then, using the passive-to-active switch, Al performed A/B tests—without and with DSP—playing identical snippets of orchestral music streamed from Qobuz.

What happened was that with the crossover in passive mode, the Maggies reached maximum excursion, bottoming out on the loudest orchestral passages and causing the sound to get jumbled due to leakage of low frequencies into the midrange panels. But the same speaker with DSP had no such problem. The bass was louder, but it didn’t affect the midrange. Not a single visitor among us said that passive had sounded better. Even The Absolute Sound magazine was impressed, awarding the Danville Signal room a “Best of Show”.

Another use of Danville’s DSP products is in phono stages to implement the RIAA curve. And it works with high level MC cartridges to achieve a signal-to-noise ratio greater than 125 dB. It turns out, as well, that DSP is particularly good at eliminating those annoying pops and clicks from vinyl recordings without harming the rest of the audio signal.

Danville’s customers are typically loudspeaker manufacturers. Al says you need to start with a good speaker, though. DSP is not a magic bullet that fixes everything. Programing of the dspNexus is accomplished using a software called Audio Weaver. Made by DSP Concepts, it’s designed to let a non-programmer perform DSP.  Danville has videos on its website that shows how to use it.

Regarding DSP’s place in our hobby, Al says that DSP is causing some skeptical “Spock eyebrows” to be raised in the industry, but he believes that in the not-too-distant future the technology will be universally accepted as a means of providing a new level of audio quality without the drawbacks or side effects that can arise in the analogue domain. In other words, Al thinks DSP has a bright future in audio. Acceptance for DSP in monitors is already widespread in professional audio. As for consumer audio, Al says, “It’s just a matter of time.”

Part 2.