Hi there, it’s indeed very cpu intense. Especially, as each source has to be encoded with their reflections.
I wrote a VST plug-in called RoomEncoder, which renders a source in a shoebox-room, and adds reflections which are also filtered depending on the wall absorption and image source degree. Source and listener can be freely placed within the room, and the source can also have a directivity which can be frequency dependent. The output is higher order Ambisonics. So to be played back, all sources for one listener have to be summed up, rotated (head-tracking), and binaurally decoded. Several instances of the plug-in can be linked, so if you change the room properties in one of them, all of them change. The plug-in renders up to 236 reflections, however, a hand full (or two) of them are enough to give a convincing room impression. Especially when combined with a FDN network. The good thing is, that you’ll need only one FDN network for all listeners and sources, so at least this one is not so cpu demanding. The FdnReverb plug-in also has a fade-in feature, which helps to not get in the way of the early reflections of the RoomEncoder. We used this setup in an interactive audio game, tracked with an optical tracking system, logic implemented in PD and rendered with Reaper. Both plug-ins can be found in the IEM Plug-in Suite: https://plugins.iem.at <https://plugins.iem.at/>. As they are open-source you could compile them yourself, to get the most out of your CPU architecture you are using e.g. AVX512 SIMD extensions. Best Daniel > Am 29.09.2019 um 14:06 schrieb Dave Hunt <davehuntau...@btinternet.com>: > > Hi Michelle, > > I believe that what you want to do is possible, but not easy. > > It is possible to move the listener in a totally synthetic ambisonic sound > field. You have to build in “distance modelling’, as well as differing > direction of each source, as the listener moves. Adding room simulation or > reverberation brings an extra layer of complexity, as the nature of the > reflected, delayed and diffused sound from each source is different at every > listening position. > > This soon becomes rather processor intensive. I have made Max patches that > take this approach, and they do “work”, but there are problems. Although > basic ambisonic source encoding is mathematically relatively simple multiple > sources, with multiple reflections, each of which have to be encoded, becomes > appreciably more involved. Moving sources require multiple constant > recalculation, preferably at near audio sampling rate. Even with just first > order ambisonics, this gets pretty demanding to do well. > > For what you are proposing, you would have to deliver a unique audio stream > to each pair of headphones, binaurally encoded to match the position of the > headphones in the “room”. Thus you need spatial tracking of each pair, as > well as the head movement data for each. This could control the ambisonic > encoding and decoding to binaural of each individual headphone signal. > > For one listener this already involves a lot of data and processing, and at > least two wireless transmission channels. For several listeners the > technology and resources required becomes uneconomic. Currently you would > probably require a computer for each listener, or possibly a very powerful > computer or two. Then a lot of programming, engineering and material expense. > > Perhaps you would be better considering a loudspeaker based approach ?? > > For this it may also be better to consider an amplitude/delay based approach > (Delta stereophony, or basic wave field synthesis), rather than ambisonics.. > This is what it appears TiMax, L-ISA, d&b’s Soundscape, and Astro and other > similar systems are based on. Again, not easy to do well. Not perfect for > everything, but then no algorithm is. > > > Ciao, > > Dave Hunt > > >> On 28 Sep 2019, at 17:00, sursound-requ...@music.vt.edu wrote >> >> From: Michelle Irving <michelle.irv...@soleilsound.com> >> Subject: [Sursound] Ambisonic Audio - Interactive Installation >> Date: 27 September 2019 at 18:22:26 BST >> To: sursound@music.vt.edu >> >> >> Hi, >> >> I'm working with an artist who wants to explore Ambisonic Audio >> and use the Audeze Mobius headphones in an audio installation. >> The soundscape will consist of recordings of various individual vocals >> spatialized >> throughout the "room". There is a video projection overhead. Hard sync is >> not required. >> >> Questions: >> 1.Is it possible to exploit the headtracking of the Mobius headphones to >> give each person and individualized experience of the audio composition. >> ie. Person A is in the far left front corner and hearing a particular voice >> in close proximity while Person B is in the far back right corner barely >> hearing what Person A is hearing? >> >> 2.If the Answer to 1. is YES - would you recommend using Max/Msp or Arduino >> for configuring hte individual playbacks (mappings between headphones and >> some sort of player) >> >> 3.I've looked at the Waves NX toolkit and I don't see a feature to >> determine virtual room size?Am I missing something or is there other tech >> that could allow me to map the headtracker to a specific roomsize? >> >> 4.Open to better ideas how to achieve an interactive Ambisonic audio >> soundscape that works with multiple headsets. >> >> thanks! >> Michelle >> >> -- >> >> Michelle Irving >> >> Post-Audio Supervisor >> >> 416-500-1631 >> >> 507 King St. East >> >> Toronto, Ontario >> >> www.soleilsound.com > > _______________________________________________ > Sursound mailing list > Sursound@music.vt.edu > https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit > account or options, view archives and so on. -------------- next part -------------- An HTML attachment was scrubbed... URL: <https://mail.music.vt.edu/mailman/private/sursound/attachments/20190929/306c80b8/attachment.html> _______________________________________________ Sursound mailing list Sursound@music.vt.edu https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit account or options, view archives and so on.
