[sdiy] Ray tracing hardware for audio simulation

[Mike Bryant]

> Or is there no "all analogue" path with zero latency in today's modern digital world!?

Definitely not !   How could you send data through the Ethernet snake from the stage to the mixer, then back again to the monitor earpieces without going digital ?  There are one or two analogue-only studios left, but almost all remaining analogue live work went digital when the X32 was launched.  There's even a couple of notepad style digital mixers now at not much more than the all-analogue ones for sole pub artistes.  Indeed for these people, they are more and more using digital microphones as they are now of good enough quality and easier to set up.

ADC and DAC delay is the primary latency in the system, and you choose a mixer with the lowest delay which is the main reason to go 96kHz or 192kHz for live - you can't hear any difference but the signal gets there faster.  
You try to keep any other delays - the mix bus, Ethernet, effects sends and DSP units at as low as possible a delay as possible, otherwise when you mix it in with the clean signal you can get all sorts of phasing problems.  

Another problem for bands is the bass player sets the timing, the drummer follows this from his earpiece, then the bass player hears the drums back through his earpiece doubly delayed so he instinctively slows down for the drummer to catch up.

Also bear in mind in live work that 'clean' signal may be some leakage into someone else's microphone.  If you have a church choir with 50 up to 100 or more live Countryman mics that are fairly omnidirectional and all at similar sensitivities, it can get quite complicated.  Live mixing isn't simple nowadays - it was much simpler in the analogue days when microphones were on stands and had cardioid responses with several people singing into one mic.

The latency issue is also why live sound autotunes don't use SFFTs or wavelets, whereas studio ones usually do as they are more accurate at the expense of far more delay.  Have a listen to the ones used in theatre musicals and they can be highly irritating to say the least, and personally I'd rather they weren't used  and just accept the average musical singer isn't Freddie Mercury or one of the Three Tenors.  A Jesus Christ Superstar production of some years ago was infamous for this.



-----Original Message-----
From: rburnett at richieburnett.co.uk [mailto:rburnett at richieburnett.co.uk] 
Sent: 01 August 2022 21:43
To: mskala at northcoastsynthesis.com
Cc: Mike Bryant; synth-diy at synth-diy.org
Subject: Re: [sdiy] Ray tracing hardware for audio simulation

On 2022-08-01 14:38, Matthew Skala via Synth-diy wrote:
> On Mon, 1 Aug 2022, Mike Bryant wrote:
>> 1mS delay is too much for live voice processing so they need to work 
>> on
> 
> Move the speaker a foot closer to the listener.

Yeah.  I'm very sceptical of claims about such small latencies mattering.  As you say it's equivalent of the monitor speaker being 1 ft further away.  Many audio CODECs have considerable delay due to the digital filters in their ADCs and DACs anyway so that sets the limit on what even a basic talk-through DSP program could achieve.

The only time I can see 1ms latency mattering is possibly for something like headphone monitoring where a vocalist might be put off by hearing their own voice coming back in the cans with a delay and mixing with what they hear directly in their head (via bone conduction?)  But then how small a delay *is* tolerable without tonal colouration?  And surely for something like "comfort reverb" on a vocal monitor mix, only the "wet" reverb part would go through the DSP system and incur latency.  
The dry vocal signal would still be all analogue, and therefore have no delay?  (Or is there no "all analogue" path with zero latency in today's modern digital world!?)

-Richie,