[revbobuk]

In a conventional auditorium, delay stacks are set so that the signal to them is delayed to match the propagation delay from the stage, more or less. That way, the sound you hear direct and the sound from each set of speakers are in step.

But often, sound reinforcement in a highly reverberant location like a cathedral is done differently, with large numbers of column speakers (to minimise vertical dispersion) operating at a very low level, so that there is minimal overlap between systems. And the 'direct' sound is practically nil - so that you only ever really hear either the speaker you are next to, or the overall reverberant field.

Intuition would seem to indicate that in this kind of situation, delays shouldn't be used - but is that correct? And that may be OK for speech, but what about for music, where the original instrument may indeed be loud enough to be heard directly?

Are there good reference works on sound in highly reverberant locations? And while something like EASE might indicate coverage in this kind of environment, does it generate information about the time coherence of multi-speaker setups?

I'm asking particularly with cathedrals and similar in mind, as I may have to prepare some reports, and could do with some more background knowledge!

[Chris L]

Hi

The Yamaha Sound Reinforcement Manual has a good section on delays and how the acoustics of the room affects how you could approach delays if you can get hold of a copy.
Ideally, you should always be able to hear a speaker directly louder than the reverberation.

I tend to delay delays to slightly behind the main PA so then your ear is tricked into thinking the sound from the delays actually comes from the direction of the main PA - but this will depend a lot on the scale of the system and the local acoustics.

So my best advice in a highly reverberent acoustic would be to use lots of sources distributed so that the audience can always hear more direct that reflected sound. It might well be that you don't mind where the sound is perceived to be coming from so you don't need to delay. If you get the chance to listen to the venue or a similar situation, then try different things out and go with what sounds clearest!
You might be able to judge the clarity and where the points are that the reverberation is louder than the source by going to a building like this with a friend. Get your friend to speak out in different locations and try to find the places where the clarity drops - you could use your friend to test out different possible locations for speakers

HTH
Chris

[Stuart91]

Is there any advantage in NOT having the speakers delayed?

[Chris L]

Stuart91, on 23 April 2012 - 11:35 PM, said:

Is there any advantage in NOT having the speakers delayed?

Fewer system controller outputs required, possibly fewer amplifiers required - particularly in a 70/100V installation... etc.

Chris

[Stuart91]

Chris L, on 23 April 2012 - 11:46 PM, said:

Fewer system controller outputs required, possibly fewer amplifiers required - particularly in a 70/100V installation... etc.

Fair point, although I was more thinking from a sound quality point of view - would it sound any better? My hunch is that it wouldn't.

[Simon Lewis]

I think this is going to depend upon what is heard at any particular location, and how much energy is being put into the reverberant sound field (assuming that it is diffuse).

If the direct sound and the local loudspeaker can be heard at comparable levels at a given location, then conventional wisdom says that the local speaker should be delayed to account for the time of flight. The textbooks will suggest additional delay for Precedence and Haas effects, but the suggestion of most practitioners is now to account only for TOF.

If the direct sound cannot be heard, and the local speaker is the only source heard (apart from the reverberant sound field) there might be the argument that putting multiple delayed versions of speaker output into the RSF would be deleterious to intelligibility, but it would be hard to determine what contribution was from the various loudspeakers.


Some of this is alluded to in the chapter on Designing for Speech Intelligibility in Handbook for Sound Engineers (Ballou) 3rd edition. That chapter is written by Peter Mapp and is (in my opinion) a better treatment than the earlier editions.

If you model with EASE it is possible to determine such meaurements as speaker overlap, arrival times and STI, so yes - it will give some information..


Can you share what the reports need to cover? Is it simply a question of delay or not delayed, or is it more to do with achieving guaranteed levels of intelligibility (e.g. maintaining 0.5 STIPA in all agreed listening positions)?


Simon

[revbobuk]

Simon Lewis, on 24 April 2012 - 01:17 AM, said:

If the direct sound cannot be heard, and the local speaker is the only source heard (apart from the reverberant sound field) there might be the argument that putting multiple delayed versions of speaker output into the RSF would be deleterious to intelligibility, but it would be hard to determine what contribution was from the various loudspeakers.

Well put, Simon. I've read the Yamaha manual (many times, over the years!) and it doesn't address this really. The report just needs to be an independent evaluation of a number of quotes for Sound Reinforcement, and I'm happy with most of the aspects of that - this is just a question that has been raised for me by the nature of the venue, and I'd like to get a handle on the theory.

I'll get hold of a copy of the Ballou book - is it '3rd edition only', or '3rd or later' that you are recommending?

It's not a question specifically of a particular STI measure - it isn't about evacuation procedures, or anything like that - but quality and intelligibilty are the key issues. After all - how are you going to measure 'time of flight' in a situation where the flight path just isn't obvious. It'll clearly have a minimum value, but that might not be the loudest signal. And how important is the direct path when you can't see the sound source anyway?

[Simon Lewis]

Bob,

Ballou's book is into the 4th edition - I ought to sell my 3rd edition copy on;-) Peter Mapp is still author for the chapter on Speech Intelligibility, and it's probably one of the best to read. There's a more laid back account on the Meyer Sound website, some good material on the MC^2 acoustic consultant's site, and - of course - academic papers by such people as Houtgast, Steeneken, Mapp, Barnett, Peutz etc.

For Cathedral systems, there can be value in progressive delay, as it usually also allows more local control of level and equalisation. There are some very nice tools in the digital directivity / beamsteering field but these are at the higher end of the price range at present. We have measured excellent results in a large reverberant space (mid band RT60 = 3.5s, STIPA = or > 0.5 over 35m) with such systems.

If you want to pop up the road to Derby for a coffee, chat and borrow a few text books, then drop me a PM. I suspect that you might be carrying out DAC advisor work? If so, I would certainly value a discussion about this.

Regards,


Simon Lewis