Thanks for the message Fons, yes that was me - I must have accidentally let that earlier email fly from the wrong account.
I think my problems with spat3d were or are in relation to the room-configuration - have since followed Jorn's suggestion and combined the early reflection modelling of spat3d with a convolution reverb tail (one of your IRs with jconvolver) - which at the moment I am sending a low-level unmodulated feed from the source. Have toned down the reflectivity of the modelled surfaces for the early reflections (there was a comb-filter effect before with the small room size that i'm using) and it's sounding much nicer. Thanks for the tip with cross-fading between ER IRs - I'll try this. Need to create some IRs too... Re. the voice - yes, distance is important - at times the voice will be heard from a position of several meters away to the rear-left - at other times, at the two sides of the seated listener and from close behind - and everything in between. The listener will have a visual reference for the source and the sound of the voice needs to match as well as possible. A question about creating IR tails: is this just a matter of trimming some milliseconds off the beginning of a complete IR with an appropriate envelope? How many msecs? Just the rising part of the impulse? Good luck finding a new home! Cheers, Iain On Sat, 2011-06-04 at 22:02 +0000, Fons Adriaensen wrote: > On Fri, Jun 03, 2011 at 09:11:35AM -0300, Iain Mott wrote: > > > I would like to know what jack compatible systems are available on Linux > > to spatialise a moving mono source in realtime with ambisonics > > (including distance attenuation) that also provide some reverberation > > cues. > > > > I've tried spat3d in csound with a GUI - and while this spatialises the > > source very well, I've not been so happy with the sound of simulated > > room reflections - perhaps it's the way I am setting it up. I'm also > > looking at the SoundScape Renderer (SSR) which is very nice - but > > doesn't seem to have an implementation of reverberation for moving > > sources. > > > > Is convolution reverb a possibility for a moving source - eg. is it > > possible to cross-fade between the inputs of ambisonic IRs set up in > > jconvolver recorded at various angular locations? This would be ideal > > for my application which involves the spatialisation of an actress' > > voice within a room - so that she appears to be walking (and talking) > > through the space in relation to a seated listener. > > I assume you are AKA 'acousmetre' ? > > Sorry for the delay, I've been quite busy the last week - having > to move home at very short notice (within 2 weeks actually, and > I haven't found a new one yet ...) > > Using a set of different ER with the same reverb tail works well > with static sources, but I've never used it with moving ones. > I've no idea if crossfading ERs for a moving source will work. > If you want to try it, the way to do it is use e.g. 3rd order > AMB panning for the reverb send, decode it to e.g. an octagon > and use the decoder outputs to drive the inputs of the convolver. > > OTOH I'm surprised about your results with spat3d, it's assumed > to be one of the most advanced tools available. > > For your project, how important is the _distance_ of the virtual > source w.r.t. the listener ? Is the actress supposed to move > very close ? > > Ciao, > _______________________________________________ Sursound mailing list Sursound@music.vt.edu https://mail.music.vt.edu/mailman/listinfo/sursound
