The future of surround sound from MIT

The future of surround sound from MIT

The Opera of the Future group is an MIT Media Lab Research Group led by Society of Sound Fellow Tod Machover. It explores concepts and techniques to help advance the future of musical composition, performance, learning, and expression. Through the design of new interfaces for both professional virtuosi and amateur music-lovers, the development of new techniques for interpreting and mapping expressive gesture, and the application of these technologies to innovative compositions and experiences, the group seeks to enhance music as a performance art, and to develop its transformative power as counterpoint to our everyday lives. The scope of our research includes musical instrument design, concepts for new performance spaces, interactive touring and permanent installations, and “music toys.” It ranges from extensions of traditional forms to radical departures, such as the Brain Opera, Toy Symphony and Death and the Powers.

Ambisonics is one of its current research projects, and here Ben Bloomberg explains a concept that could very well be the future of surround sound….


Anyone who has gone through the process of designing a home theater understands the difficulties associated with locating speakers. Unless the theater is designed and built from the ground up, it is seldom possible to put speakers in perfectly ideal positions. Ambisonic encoding provides an abstraction layer, which allows one to place speakers independently of sound. It is an elegant method of representing audio such that the locations of ‘virtual sources’ are represented in the most accurate way given any possible speaker configuration.

At Opera of the Future, we strive to create spectacular next generation live performance systems that are realistically usable in harsh environments such as touring productions. Systems must be flexible enough to fit into any venue, ranging from small offices to concert halls, and extremely simple to setup and configure. Using Ambisonic encoding, it is possible to arrive at a venue, place surround sound speakers wherever it is most convenient and have a fantastic, consistent surround sound field.

Although Ambisonic encoding is a well-known technique for reproducing pre-recorded content, it is not often used as live tool because latency associated with encoding and decoding is too great. For our latest production, Death and the Powers, we have developed an AppleCoreAudio AudioUnit DSP engine, which is capable of processing 128 channels of ambisonics at a staggering 32 samples of delay. This allows us to take ambisonics to new heights in live entertainment and sound reinforcement.

Existing theatrical surround sound processors cost upwards of $120,000. This system can run on hardware ranging from a small FPGA-based 8-channel system ($60) to a 128 channel MacPro based system ($5,000). It can be automated using industry standard OSC and uses any CoreAudio compatible audio interface for audio I/O.

How does it work?

Ambisonic encoding uses 3rd order spherical harmonics to represent virtual sources with 16 directional coordinates that function equivalently to X, Y and Z coordinates. Using a 16 axes, instead of 3, to specify the location of a virtual source allows greater “perceived resolution” in the surround sound field. Thus adding more speakers can produce a more detailed result rather than a ‘blurrier’ field.

Bowers & Wilkins to the Rescue!

We have been developing ambisonic systems at Opera of the Future for 3 years now. Until the present, we were unable to find a speaker that was small enough and sufficiently detailed to justify use in large quantities for testing in our lab and studios. It should be mentioned that we spend our days using our 800D and 805s speakers in the studio for production work, so we were extremely picky when it came to finding a speaker that would work well for our large-scale ambisonic tests.

Just when production for Death and the Powers was beginning to reach its peak, Bowers & Wilkins introduced us to the M-1, a four inch speaker with a one inch aluminum nautilus tweeter. Needless to say, it seemed like the perfect speaker for us to use. In a flurry of emails and phone calls, we managed to obtain 24(!) of them for some testing and experimentation.

The results could not be more positive. After some time testing and tweaking decoder weightings and crossover points, we found that paired with two of our 800Dʼs and an ASW-855 subwoofer, the system sounds incredible!

The first debut of the system was during the open house for the grand opening of our new building, but it has since been the centerpiece of many demos, including one for Gustavo Dudamel, conductor of the LA Philharmonic.

The System

The DSP rig consists of a MacPro loaded with AULab and MOTU Digital Performer running in real-time mode (a fantastic resource) and a MADI-based Solid State Logic 128 Channel PCI- express card. The system clock is generated on board a separate SSL AlphaLink SX and sent over MADI fibre-optic cable to the computer. The DSP runs at 96Khz and the converters are 24-bit.

The M-1s are spaced evenly at 15 degree increments with two 800Ds in front at 30 degrees and an ASW-855 Subwoofer in the rear. The M-1s have a 1st order high pass filter at 140Hz and the subwoofer is crossed over at 100Hz and 180 degrees out of phase. The 800Ds run full range.

System Specs:

MacPro 8-core 2.26Ghz w/6GB RAM SSL MadiXtreme128 SSL Alpha-link MADI SX MOTU Digital Performer 7.1

Apple AULab 2.1 (19) B&W M1 Speakers (2) B&W 800D Speakers (1) B&W ASW-855 Subwoofer (2) Rotel 1512 Amplifier (2) Rotel 1091 Amplifier.

Whatʼs next?

The M-1s will play an important role in Death and the Powers as part of a 128 channel surround-sound system! Stay tuned for more information on the design and implementation of that system.

Ben Bloomberg is an undergraduate at the Massachusetts Institute of Technology who has been obsessed with audio since age 9. Currently he designs surround sound infrastructure for live entertainment at the Media Lab. His systems have been featured internationally in productions such as the world premier of Tod Machover’s Skellig, and more recently, Death and the Powers at the American Repertory Theater in Cambridge, Massachusetts.

Death and the Powers will premier for the public at the Salle Garnier Opera House in Monaco on September 24, 2010, and then tour to the United States in the spring of 2011. For more information on the production, visit


  • Tito Zenteno says:

    Wow, it´s perfect but the price is big, I waiting for this tecnology, congratulations.

  • Andrew says:

    Amisonics isn’t a new technology, far from it, but I am glad to see that b&w are getting involved in the action. Research facilities in England and Scotland, as well as many in Europe are striving to replace the ‘fake’ surround sound with ambisonics

    as a side note, smaller speakers seem to work better for ambisonic arrays over larger counter parts – shame the price is still as high for 16!

  • Mozzy says:

    I question the ethics behind the crossover points, it appears a bit of a mess, with the sub rolling off at 100Hz, the M-1’s rolling off at 140Hz, and the 800’s runnig full range, not a very uniform bass coverage.

    Setting all speakers to roll-of at the 80Hz point to the subwoofer would provide a far superior result.

  • Ben says:


    Very few pro-audio system designers set their crossover frequencies to be the same between satellite speakers and subwoofers. Keep in mind that with a 1st order HPF or LPF, the fall-off is 6db per octave (12db for a 2nd order and so on..). That means that if both speakers are set to the same filter frequency, you actually end up with a pretty large excess of low end where the filters overlap. For that reason, it’s always smart to allow a little breathing room so that the slopes between the high-pass and low-pass filters crossover in a relatively flat way. The best way to do this is by putting the system on a scope and looking at its spectral response. We do this often with very large sound systems and so we’ve gotten pretty good at doing it by ear/educated guessing. In the end, it’s really about personal taste.

    I set the HPF on the M1s up higher than usual because anything below 150Hz doesn’t localize very well in surround sound. A lot of our material has pretty significant 35Hz to 125Hz. By keeping that out of the M1s, we get more volume in the high-end. The Nautilus tweeters are fantastic at projecting and the 4″ drivers are great at mid-range, but for our systems we like to make sure they have lots of support underneath so we’re not depending on a 4″ driver to get bass 20 feet across the room.

    One reason we have a sub at the back of the room is to spread out the bass field. We found that with the sub in the rear combined with the full range low-end of the 800Ds, the low material is sufficiently “around” so that high-end cues from the M1s make the bass feel like it’s coming from one place or the other.

    Of course, the best way to design a system is to model it and the room together in a program like EASE! Then take a long listen and make changes based on your ears. Numbers are great… but in the end what really matters is what you hear.

  • Hal Owen says:

    Speaking as a devotee of home surround audio, (aka five channel SACD/no video/non pro listener,) I’m suprised to see such an elaborate speaker set up with only one sub. Certainly the 800s have a georgous full frequency balance but it has been my experience to notice a low end roll off starting around 40 hz. Can one sub adequately handle all that valuable information below say 35hz or does the effect of such a full sound field suffice? In fairness, I should also say that one audio bench mark for me is the CBS/Sony multi channel recording with organist E.Power Biggs performing J.S.Bach on “The Four Antiphonal Organs of the Cathedral of Freiburg”, (SS87983.) Recorded in October 1973, here is a remarkable document demonstrating the sonic glories of surround audio in an ever so reverberant, (6 second decay time I believe,) one hundred meter long structure. Having visited the Freiburg, Germany Cathedral, I can say the live listening experience was for me, cathartic to say the least – home listening to this recording is almost as satisfying.
    Best wishes, Hal Owen

  • Cielspacing says:

    The low bass isuue could be bettered with, say 6″, 8″ or 10″ for the 16 mid drivers, which will allow for those rather small speakers, to reproduce down to 100 kHz or even 80 kHz with more ease than the actual 4″ drivers.
    Concurrently I would place the Sub at the central location, pointing upwards, this way ensuring a better 180° coverage, assuming the audience are going to occupy the place inside the 16 speakers’ circle.
    This configuration would be the optimus for the projection of music composed and recorded multichannel.

    As it is, (in the picture) the system looks like the perfect setup to monitor, mix and control the recording… at the front position. That would explain the inclusion of a 800D pair, other than to utilize the system for regular stereo centered music material.

Add a comment

We welcome debate within Society of Sound, but please keep it friendly, respectful and relevant. We have a few house rules which we ask you to abide by to keep the debate intelligent. Read more.
Product enquiry or support issue? Please click here.

Related Posts

MIT Media Lab 25th Anniversary Event

On October 25th at 7pm, Opera of the Future Group at the MIT Media Lab will screen an HD recording of the premiere of Death and the …

Tod Machover’s robotic opera premieres in Monte Carlo

The groundbreaking robotic opera Death and the Powers, created by Society of Sound Fellow  and MIT Professor Tod Machover, premieres …

Tod Machover’s ambisonic sound system

Tod Machover, Professor of Music and Media at MIT and Fellow of Society of Sound has recently been experimenting with a 3D Ambisonic …