Re: Limit in sample or ms for circular routing (Re: Minutes of Audio WG Teleconference, 2012-05-30)

On Thu, 31 May 2012 16:29:26 +0200, Olivier Thereaux  
<olivier.thereaux@bbc.co.uk> wrote:

> On 31/05/2012 12:28, Robert O'Callahan wrote:
>> I think 128 samples is a bit unfortunate for authors since they specify
>> the time in seconds. Couldn't we make this a limit in seconds?
>
> This was a large part of our discussion on the call:
> http://www.w3.org/2012/05/30-audio-minutes.html
>
> Chris seemed to prefer the idea of a limit in seconds, Philip seemed to
> expect it would make implementations harder.
>
> Others will be able to say this with more authority, but my
> understanding is:
>
> Seconds
> * Easier for authors
> * Will sound the same in all implementations regardless of sample rate
> * May run into issue where limit is too low depending on
> implementation's sample block size
> * implementation would probably have to clamp to the closest "block"
> size anyway
>
> Samples
> * Easier to implement, less risky
> * May sound slightly different depending on implementation, but should
> not be noticeable

To be specific, if one tries to create a loop with 0 delay, it will be  
clamped up either to a fixed number of samples (probably 64 or 128, to be  
defined) or to a fixed number of milliseconds. Defining it in milliseconds  
breaks for low sample rates, e.g. a 3 ms clamp at 8 KHz is just 24  
samples, below the block size any implementation will want to use.

The problem with defining it in terms of samples is that if the graph is  
allowed to run at the native sampling rate, then clamped delay (in ms)  
will be different for different machines.

(Another approach is to just not allow loops, and I'm looking forward to  
hear what Robert O'Callahan has to say on that.)

-- 
Philip Jägenstedt
Core Developer
Opera Software

Received on Thursday, 31 May 2012 16:04:07 UTC