On Location with Audyssey Laboratories

Nearly two years ago we were introduced to a revolutionary room equalization system called Audyssey which was soon to be debuted in the up and coming AVR-5805 receiver flagship from Denon. Prior to reviewing this new super receiver, we sent one of our freelance writers, Patrick Hart , to Audyssey Labs for a first hand demo of their new system.

See: Audyssey Labs' MultEQ Report

With such a favorable report, we were most eager in getting this system into our own test labs. Along the way we discovered a few glitches in its execution, including its lack of correction below 100Hz. Both Denon and Audyssey were proactive in resolving this issue and shortly a new improved firmware version was released which made its way into all Denon Audyssey enabled receivers. Our detailed test report of the AVR-3806 and AVR-5805 will follow shortly.

We were impressed with Audyssey's commitment to improving their technology and when we discovered they were working on a new stand alone processor, among other things, we were most eager to arrange an On Location Tour of their facilities and get the scoop.

When we arrived, we were greeted by Samit Varma who is now the Director of Operations for Audyssey. Samit is no stranger to acoustics, having worked on nuclear subs in the Navy as a Weapons Officer on the USS Nebraska. Dealing with underwater acoustics is an incredibly effective way to hone your skills for a career in the entertainment theater industry. Samit also taught engineering at USC to ROTC midshipmen and got his Masters (MBA) there where he connected with the budding Audyssey. He came on board with the company just a few months after Michael Solomon took the position of CEO in December of 2004.

Company History

The history and progression of Audyssey is interesting in that we are witnessing a company birthed out of a need to improve the status quo of what is largely an overlookied issue in home theater today - digital correction of loudspeaker output to achieve better overall in-room performance. Add the fact that the company has grown from a single software-only product into a new line of hardware and licensing products and you're witnessing the beginnings of something with enormous growth potential.

Denon was Audyssey's first official licensee customer with the introduction of the AVR-5805 "super-receiver". It hit a number of firsts, not the least of which was the integration of Audyssey's MultEQ XT system and, at the time, what we considered to be the first promising and usable room correction system integrated into an AV receiver. What makes Audyssey unique is that it attempts to remove things that aren't supposed to be in the loudspeaker-to-listeners signal path. Audyssey doesn't set out to be an "effect" company - they don't seek to add anything to the sound.

Audyssey started out with just one employee and is now up to 18. As a company, Audyssey has now grown into several licensees, with new hardware products (yes, we said hardware ) and lots on the horizon.

The founders are Tomlinson Holman (Chief Scientist and the original developer of the THX sound system), Chris Kyriakakis (Chief Technology Officer), Phil Hilmes (VP of Engineering), and Sunil Bharitkar (VP of Research and Development).

Biographies of Key Audyssey Personnel

Dr. Sunil Bharitkar, Ph.D., Vice President of Research/Founder
Dr. Sunil Bharitkar, Ph.D. has published over 50 technical papers on digital equalization of room acoustics and audio signal processing, including his award-winning paper on the influence of reverberation on multi-position equalization performance. He co-authored the textbook Immersive Audio Signal Processing, published by Springer. His room equalization and signal cancellation research, at both USC and Audyssey, has resulted in several patented or patent-pending co-inventions. He is a frequent reviewer and lecturer for various IEEE journals and conferences, the Journal of the Acoustical Society of America, and Signal Processing from EURASIP. His research background includes signal processing, pattern recognition, acoustics, speech & audio processing, and auditory perception.

Philip Hilmes, Vice President of Engineering/Founder
Philip Hilmes' expertise in systems engineering brought him to head Audyssey's effort implementing breakthrough audio technologies into an ever-increasing range of professional and consumer audio products. Hilmes has held engineering positions in communications & digital entertainment at DIRECTV, where he created technology increasing the amount of data transmittable through limited bandwidth, thus augmenting the number of channels that can be transmitted through existing broadcast satellites. He has three pending patents.
Prof. Tomlinson Holman, Chief Scientist/Founder
With over 35 years of audio industry experience, Prof. Tomlinson Holman served as Chief Engineer of Henry Kloss' Advent Corporation, founded Apt Corporation (maker of the Apt/Holman Preamplifier), and has written three books on audio. For his work at Lucasfilm Ltd in developing the THX Sound System, Holman won an Academy Award. He is currently professor of film sound at the USC School of Cinema & Television.
Prof. Chris Kyriakakis , Chief Technology Officer/Founder
Prof. Chris Kyriakakis is Associate Professor of Audio Signal Processing at the University of Southern California 's Viterbi School of Engineering. He is founder and Director of USC Immersive Audio Laboratory, and Deputy Director of the Integrated Media Systems Center , a National Science Foundation research center. He has authored two books and more than 100 journal and conference publications. His expertise is in the areas of immersive audio signal processing, microphone arrays for robust sound localization, multichannel audio streaming over high bandwidth networks, virtual microphones for multichannel audio synthesis, high fidelity multichannel audio coding, and multiple listener room equalization.

Michael Solomon, Chief Executive Officer
Michael Solomon built his career over the past 30 years as an entrepreneur in the audio and computer industries, a software and computer industry executive, and a venture capitalist. He worked with Apple computer through his own business to set up Apple's early dealer network in the early 1980's, was the original sales and marketing VP of Aldus, the creators of PageMaker software which spawned the desktop publishing industry, worked in Venture Capital as a partner at Mohr, Davidow Ventures during the late 1990's, and served on the Boards of several successful companies as well as being CEO or founder of three other ventures along the way. He joined Audyssey in December 2004 as CEO, thereby, connecting the dots in his career to his first job as a retail audio sales person while in school.

Samit Varma Director of Operations
Samit Varma holds a Bachelor's in Mechanical Engineering from the United States Naval Academy . He further studied Nuclear Engineering from the Naval Nuclear Power Program. After serving as an officer on the Ballistic Missile Submarine USS Nebraska, Samit earned his MBA from the University of Southern California where he was discovered by the Audyssey team and signed on to bring the leadership, discipline and focus he learned as one of the Navy's top submarine officers.

USC
The relationship between Audyssey and USC (the University of Southern California ) is a symbiotic one and really impressed upon us the unique position Audyssey enjoys with respect to resources and capabilities. Chris & Tom are both active professors with research facilities at USC. To say that this relationship works well for Audyssey would be a tremendous understatement.

Publications and Awards

Audyssey has received much acclaim in the industry to date. From award winning papers on the influence of reverberation in multi-position equalization performance to having over 50 technical papers published on the topic of digital equalization of room acoustics and audio signal processing, Audyssey has been invited to give talks and tutorials at the International Conference on Acoustics, Speech, and Signal Processing (ICASSP) and has presented several papers in the Audio Engineering Society Conference, the Acoustical Society, and other signal processing conferences. Sunil and Chris have published several peer-reviewed journal papers and a book titled Immersive Audio Signal Processing which is available from Amazon.com.

Research Facilities

The Immersive Audio Laboratory Audyssey uses at USC is truly amazing. This sound room consists of an acoustically controlled environment consisting of close to $30k in absorption and diffusion treatments. The room has 14' high ceilings in order to take full advantage of a new surround format that Tom and Chris have been working on (more on this later). They went even further by building the room from the ground up using innovative construction measures to ensure total sound isolation from the outside world. This included having a custom sound isolation door costing a whopping $10k that had the acoustical properties needed to preserve the noise isolation of the room.

What was most interesting was that the room didn't have a dead or anechoic feeling as many improperly treated rooms often do. Instead the room seemed very vibrant and natural sounding while listening to music. Intelligibility was very good when engaging in conversations in the room.

The Immersive Audio Lab is a unique facility that is part of the Integrated Media Systems Center (IMSC), a National Science Foundation engineering research center at the USC Viterbi School of Engineering. The lab was founded in 1996 by Chris Kyriakakis and Tomlinson Holman both Professors at USC in Engineering and Cinema respectively. It moved to its current location in January, 2005. Kyriakakis and Holman designed the space so that it can be used for experiments in room acoustics, sound reproduction, and sound perception. Their requirement for 17' slab-to-slab height was met by making the entire 2nd floor of the building 3' taller than the rest of the floors!

The main lab is approximately 30' x 40' and was designed as a room-within-a-room for noise isolation. The design spec was NC-15 and through very careful planning, acoustical design, and construction the spec was actually exceeded resulting in an NC-10 facility. The walls are double gypsum board on metal studs with an inner wall spacing of 6". The floor is raised concrete (with mini-jacks on pads) to isolate from the slab underneath. The door is STC-56 to prevent leakage in both directions. Great care was taken in the design of the HVAC system that is completely isolated from the rest of the building and is constructed with extra wide ducts to minimize air flow noise.

After having addressed the noise isolation issues, the design focused on the acoustics of the room. When empty, the room had a reverberation time of 3.2 seconds. The ITU recommendation for surround sound playback in that room volume is 0.6 seconds so a combination of absorption and diffusion was used to achieve the desired acoustics. First, thick carpeting with dense padding was laid. That was followed by 15 RPG Modex Broadband absorbers along the side walls that exhibit fairly uniform absorption down to low frequencies. Early reflections on the side walls were treated with RPG Abfusers. Front wall, back wall, and ceiling reflections were treated with RPG Skylines. The reverberation time in the finished room was 0.7 seconds empty and 0.6 seconds when filled with 20 people.

Introducing the 10.2 Surround Format

Actually it isn't quite so new considering Tom has shown this 10.2 technology for years and has10 such permanent facilities that he's built for others worldwide including -one at Bjorn's Audio Video in San Antonio . The practicality of this format is somewhat limited due to logistics of integrating the additional speakers for its realization and the fact that it does require significantly more processing power. The motivation behind 10.2 was to demonstrate that the audio industry focus on higher sampling rates is not going to produce the most perceivable difference. Since the limits of human perception have already been met in word length and sampling rate, the next revolution will inevitably be in the number of audio channels to better match human spatial sound perception.

In evaluating the important improvements which could be made to standard 5.1 in order to improve the surround sound experience, the first place to fill in is to the sides. Psychoacoustics research says that side reflections are important in the perception of source width and the preferred direction for such reflections is at 55 degrees. In response to this research, 10.2 adds new width Left Wide and Right Wide speakers and channels (with a 35 ms delay to simulate the first reflection of a typical shoebox shaped concert hall). These additional tracks are added in the mixing engineering phase.

A single discrete Surround Back speaker is utilized in 10.2 to fill in the wide gap left there by the standard surround speakers at ±110 degrees. The next thing to be reproduced was the ceiling reflection above the orchestra in concert halls. A Left Height and Right Height channel were subsequently added to more accurately recreate "depth" in the soundstage.

Finally, two independent subwoofer channels are provided, each bass managed from its respective side and also available for separate left and right LFE tracks. On most tracks this will still result in a dual mono subwoofer system, but the potential for greater fidelity is there with independent Left & Right subwoofer channels. Audyssey is still conducting research into the science behind this decision. They believe that although bass waves below 80 Hz are omnidirectional, there is research evidence that shows stereo bass can improve the sense of envelopment and add a further sense of realism to the surround experience.

10.2 Speaker Layout Diagram and Explanation

The above diagram illustrates the placement of the additional speakers along with the standard 5.1 ITU setup. The channel naming is as follows:

L, R, C, LS, RS, LHeight (LH), RHeight (RH), LWide (LW), RWide (RW), Back Surround (BS), L SUB, R SUB

To the standard 5.1 layout, 10.2 adds speakers based on both acoustical and psychoacoustical considerations. Side wall reflections in concert halls, for example, have been found to add to the sense of spaciousness when implemented properly. Ando found through extensive listening tests (Y. Ando, "Concert Hall Acoustics", Springer, Berlin, 1985) that there is a preferred direction for first reflections, preferred spectrum, and preferred delay. The LW and RW speakers are intended to reproduce such reflections picked up by wide or outrigger microphones in 10.2 recordings. A second, and equally important, role for the wide speakers is that they shorten the gap between front and surround speakers. This allows more accurate placement and panning of sounds from front to back.

The second most important reflection in a concert hall, after the side, is the one from the proscenium above the stage. The Left and Right Height channels are designed to reproduce the sound from that direction. They are also appropriately processed and delayed to seamlessly blend with the direct sound from L, C, and R.

Finally, the back surround speaker is used to fill in the gap between the left and right surrounds to allow for both discrete placement and seamless surround ambience.

The Need for 10.2 over Conventional 5.1, Based on Human Perception of Sound

5.1 channels (a term coined by Tom Holman at a standards meeting in the early 1990s) was the minimum number of channels needed to reproduce a front soundstage and an ambient surround impression. It is, by no means, the limit as human perception can discern many more directions. 10.2 was designed to fill in some of the perceptual gaps that are left unfulfilled by 5.1. For example, Makous and Middlebrooks (J.C. Makous and J.C. Middlebrooks, "Two Dimensional Sound Localization by Human Listeners," J. Acoustical Soc. of America, vol. 87, no. 5, May 1990, pp. 2188-2200.) showed that human localization is much more precise in the horizontal plane and the front hemisphere and degrades for the sides and rear. That would indicate that for a fixed number of channels, more should be allocated to the front than the back. Unfortunately, the surround industry has moved in the opposite direction by adding more surround channels than front channels (e.g., in 7.1 systems). While additional surround channels can be useful in reproducing ambient sound and improve audience coverage in large rooms, current systems do not map well with the needs of human perception.

In other work by Rebscher and Theile (R. Rebscher and G. Theile, "Enlarging the Listening Area by Increasing the Number of Loudspeakers," 88th AES Convention, 1990, Preprint No. 2932.) it was shown that there is a direct relationship between the size of the useful listening area and the number of screen (front) loudspeakers. Moving from 2 to 3 and then to 5 front speakers (as found in 10.2) the area in which imaging distortion is minimized increases from a few centimeters to over 3 meters.

From R&D to Production

We sat with Sunil Bharitkar and discussed the procedure for bringing new products to market. Sunil earned his PhD in the area of room equalization. When their research was in the formative stages and finally came to fruition, Chris, Tom, Phil, and Sunil decided to start the company and license the technology. They found that their objective and subjective measurement showed that MultEQ technology was solving many of the room correction problems that parametric EQ methods had been unable to solve in the past.

Research and development is the beginning of every new concept which is worked through and then eventually productized. Sunil interacts with both Chris & Tom (and also works closely with Philip Hilmes) to develop ideas into working models that can be expanded upon. These ideas are worked into a more refined product which then led to demonstrations to other companies. One of the primary tools he uses to do research is MatLab which is integral during initial testing phases. These working models then get passed on to the DSP engineering group - where they can port a real product to the appropriate DSP platforms. Having an efficient and well thought-out chain of events is core to what Audyssey does. They shared some of this with us and it breaks down as follows:

1) Identify the Problem
Chris Kyriakakis & Tom Holman identify the problem and come up with ideas to tackle it. For example, MultEQ was created because Tom was going into theaters with stacks of equipment and taking weeks at a time to complete what was a very laborious process to calibrate a large listening room. Lots of skills were needed and the process was laborious and could not be easily replicated in different rooms. Tom believed there had to be a way to develop an automated way to handle acoustical issues since the procedures and process were essentially repeatable and defined.

2) Research Phase
This is where Sunil Bharitkar gets involved and discovers how specifically Audyssey can fix the problem. For the original MultEQ, he was able to take USC research students and devote quite a bit of manpower to the task, using the tools and software that Audyssey makes available for tackling these issues.

3) Commercialize the Product
Phil Hilmes takes the ball from here, his job primarily being to lead the team of DSP engineers and make the solution practical. This means fitting it onto a DSP chip, making sure it operates in real-time, improving system efficiency, and verifying that the solution still meets the requirements of the product or software algorithm.

4) Licensing
The final stage is where all of this R & D is recouped - namely licensing the finished product to consumer electronics companies. In the case of new hardware products, they bring them to market through the proper dealer network channels. Audyssey's goal is to work with all industries and markets: professional, consumer, automotive, and even the personal computing industry. They want to be anywhere speakers produce sound.

Product Roadmaps

Audyssey offers 5 levels of licensing of their EQ technology as follows:

MultEQ Pro - Currently runs on the Audyssey Sound Equalizer standalone processor with 16x resolution FIR filters, up to 32 room measurement positions and enhanced (up to 64x resolution) bass correction

MultEQ XT - 8x resolution FIR filters with up to 8 room measurement positions

MultEQ - 2x resolution FIR filters and up to 6 room measurement positions

Audyssey 2EQ - 1x resolution FIR filters and up to 2 room positions

Audyssey EQ - fixed system correction intended for HTIB systems and digital TVs to better linearize system response

Audyssey algorithms are currently supported by Analog Devices and Texas Instruments chipsets allowing for a wider diversity of hardware platforms manufacturers can develop their receivers or processors on. Audyssey is currently working on integrating their technologies into car audio applications and home audio subwoofer systems.

New Audyssey Sound Equalizer with MultEQ Pro Software (MSRP $2,500)

Audyssey has recently introduced a new stand alone processor featuring MultEQ technologies but with double the filter resolution of what is currently available in receivers, as well as enhanced processing for bass correction. It is an 8 channel device that connects between the A/V processor and the power amplifier and is intended for systems with separate components.

Audyssey MultEQ Pro allows up to 32 calibration points (unlike the capped 8 position calibration of MultEQ XT). The Sound Equalizer is intended for the custom installer market. An Audyssey-trained installer uses the Installer Kit that comes with an individually calibrated mic (with cal file that matching it to an industry-standard ¼" B & K reference mic), calibrated preamp, USB cable, mic stand, and 72' of XLR cables. Best of all it comes in a beautiful padded carry case that can be taken on-site for MultEQ Pro installations.

Audyssey built a pseudo anechoic chamber where they measure each mic that ships with the MultEQ Pro kit. They take that response curve and compare it to a reference B & K microphone (model #4136). They measure the mic response and invert it to create a difference curve to correct the mic so it matches the response of the multi-thousand dollar reference system. The manufacturing variance between mics is great enough to warrant them to be serialized (this is then uploaded into the software). Not only is the mic calibrated, but the preamplifier is also measured so that its consistency is maintained throughout the entire signal chain. These measures are critical to achieve the precision needed for meaningful calibration across the entire frequency range.

The MultEQ Pro algorithms allow up to -20 dB cuts, and 9 dB boosts for the entire frequency spectrum it attempts to correct. It employs a -9 dB signal reduction at the input to maintain system headroom. It is designed for industry-standard pre-processor line-out levels of 1.5 Vrms.

Surround Sound Demos

While we sat in the Immersive Audio Lab, we were given a brief history of surround sound including Tom Holman's direct involvement with the birthing of these technologies. He was the first to call discrete surround sound a "5.1 system". As most home theater enthusiasts know, 5.1 stands for Left Front, Center, Right Front, Left Surround, Right Surround, Subwoofer (a limited bandwidth channel, hence the .1). The demo discussed our previously mentioned limitations of conventional 5.1 formats which included the lack of convincing front to back panning of information. According to research, human hearing has degraded localization capability at 90 degrees. The benefit of 10.2 allows for a more naturalistic surround presentation since it provides the critical height and width data our brains are most perceptually sensitive to.

At this point, our brains were nearly saturated with digesting all of this fascinating information and were finally ready to hear some demos. Audyssey conducted some demos in conventional 5.1 and in 10.2 to compare and contrast the experience.

First up was a Diana Krall demo. She was on stage with bassist backing her while she introed her band. The sound was expansive with the walls of the room virtually disappearing. There seemed to be no disconnect between hearing her in the Immersive Audio Lab vs actually being at the live performance. The only thing missing (and thankfully so) was the usually annoying viewer(s) behind us shouting or singing along off key while Diana was singing "All or Nothing at All".

The next track was a chorus and pipe organ piece by Tom Vigneri. The venue was Methuen Hall. In 10.2 you literally felt immersed into the hall. The recording was done with 24 individual mic scattered around the stage, with some located above the chorus and positioned at two different heights in proximity of the organ. Some of the mics were used to capture hall ambience as well as direct sound from the soloists. There was also a mic position facing the wall for capturing back channel information.

In 10.2 you really had that "live" experience. A quick switch over to conventional 5.1 collapsed the front soundstage, making the listening space seem much smaller and sonically dull. We no longer had the impression of the "live" experience which our ears quickly adapted to upon instantly hearing. The leap in sound quality between 5.1 to 10.2 is comparative to first hearing a 2CH recording vs a dedicated 5.1 recording. Once you hear the latter, it's difficult to go back to the former.

What was truly impressive about this recording was its authentic rendering that we've simply never heard in any conventional 5.1 recording. Chris told us the success of this recording had largely to do with the fact that when you record and mix the tracks yourself, and know the composer, it really goes a long way in ensuring you've captured and produced an accurate multi channel rendering of the experience.

Last but not least, we heard the beautifully recorded Herbie Hancock Butterfly song which we learned consisted of 48 original tracks that they remixed especially for this project.

In 10.2, the mix had a very pinpointing and enveloping soundstage which was undoubtedly thanks to the height and width channels. A good portion of this experience was lost once we switched to conventional 5.1. The recording now took on a dryer more sterile feeling.

Wrap Up

Our visit to Audyssey and the USC Immersive Audio Lab was enlightening and refreshing. We were pleased to see the direction that this once little start-up company is now taking. They have broadened their horizons, achieving a much larger portion of the market share from entry level all the way to targeted custom installer solutions for uncompromised theater installations. Audyssey's future looks bright thanks to their very talented research and design staff, and savvy business folks who address market needs based on consumer feedback, professional feedback from installers and reviewers, and observed market trends. We look forward to exploring the Audyssey potential as it unfolds over the next few years into great new products and envelop pushing technologies.

Pictured left to right

(Clint DeBoer, Michael Solomon, Gene DellaSala, and Chris Kyriakakis )

For more information on Audyssey Laboratories, please visit www.audyssey.com .