In my last post I talked a bit about why 91Èȱ¬ R&D is interested in Ambisonics. This week I am going to talk a bit about what I've done so far, and what we might do in the future.
As most academics will tell you, the first step when undertaking a new research project is a literature review. This is to find out what other people have done already (so you can avoid re-doing work) and to help you get an idea of what it is you need to find out. There is a fair amount of literature about Ambisonics and there has been a small but dedicated research community surrounding it since it was developed over 40 years ago by .ÌýI spentÌýthe first month or so of the project reading.
I've talked a bit about working in the North Lab, and Rowan's talked about setting up a new research facility. When I arrived in Manchester to start the Ambisonics and Periphony project (before the R&D North moved into the new interim lab facility) one thing that was missing was a listening room. Kingwood Warren had an excellent ITU standard listening room, but there was no such facility in our north base. New Broadcasting House, on Oxford Road was built in the 1970s and I think it would be fair to say parts of it are past their best. Taking a short cut back from the canteen I discovered a decommissioned radio studio in a dark corridor which turned out to have, perhaps not , but good enough acoustics to set up a listening room. I set up the room with 14 speakers for the Ambisonic array; a cube (a square above and a square below) and six in a hexagon at listener head height. I also added two extra speakers in the horizontal plane so I could compare mono, stereo and 5.1 with Ambisonics.
Perhaps the most common way of making Ambisonic recordings is by using a . This is a microphone system that is capable of outputting a B-format signal, and there are a lot of high quality Soundfield recordings available online from sites such as and if you want to explore them.Ìý
One of the things we wanted to look at was how Ambisonics might be integrated into typical 91Èȱ¬ production workflows. The perfect opportunity for this came up with the recording of The Last Night of the Proms, last September. The week before the beginning of the Proms season we slung a Soundfield microphone up above the conductor's position in the Royal Albert Hall. On the last night we came back with an external soundcard and a laptop to record the B-format signal from the Soundfield mic. The Proms is recorded for the 91Èȱ¬ by a specialist outside broadcast company called . I worked with SIS to get hold of all the close microphone audio tracks of the Proms (this filled a 300 gigabyte hard drive!). l converted these signals into B-format by mathematically placing them in a soundfield and mixed them with the original signal, that we recorded to recreate the sound of the event.
In addition to the recording of live events the 91Èȱ¬ produces a lot of radio drama where the final programme is artificially created by mixing recordings of actors performing along with recordings and archive sound effects. To assess how Ambisonics might work in this type of production I made contact with a producer of Radio 4 Drama who was about to record The Wonderful Wizard of Oz. This drama was recorded in a radio drama studio (a fairly dead space) where Actors perform to aÌýstereo microphone. We set up the Soundfield microphone in the studio and the actors performed around it. The sound effects were given to me afterwards as mono and stereo audio files. I produced this in a similar way to the Proms mix, but using a combination of B format ambience and mono effects mathematically placed in the sound-field. Another thing I experimented with here was using of B format signals. Using B format impulse responses of reverberant spaces to add ambience to the dry B format recordings of the actors.
We also conducted a number of listening tests toÌýassessÌýsubjectively some of the work we did. Chris Baume, who had been working atÌýKingswood Warren on the project, did some listening tests investigating how important periphony is to the listening experience, while I focused on comparing the audience's enjoyment of Ambisonics with their enjoyment of stereo and 5.1.
That pretty much brings us up to the present. I'm currently analysing the results of the listening tests and writing up the whole project into an AES paper that Chris and I will be . While this project has shown what first order Ambisonics might do to for the 91Èȱ¬, it has also shown that there is more research to be done. To paraphrase Donald Rumsfeld, this project has turned quite a few unknown unknowns into known unknowns, and I look forward to turning them into known knowns in the future.
Thank you very much for all your comments on the previous Ambisonics research post. We've found them really useful in producing today's post so hopefully some of your questions have been answered. Please do leave comments below to continue the discussion.
Comment number 1.
At 31st Mar 2010, Bruce Wiggins wrote:Good to hear that the 91Èȱ¬ are progressing with Ambisonics. I really do believe that the speaker agnostic future proofing it provides are perfect for Broadcast, computer games etc..
For those interested, Ambisonics is currently being used by Codemasters in their Colin McRae series of games (an interview with more details can be found at )
If you're interested in actually using Ambisonics in music production, have a look at the screencasts available from
Complain about this comment (Comment number 1)
Comment number 2.
At 7th Apr 2010, Jon wrote:Hi Anthony,
As a 4th year electronic & computer engineering student at the University of Birmingham, I've recently completed a project in algorithmic reverb using ambisonic B format. Its fantastic that the 91Èȱ¬ are considering working with ambisonics and I'd love to hear more about what you're finding. It'd be great to be able to compare the convolution reverb effects you've been experimenting with to my algorithmic approach too!
Jon
Complain about this comment (Comment number 2)