[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [sc-dev] linux



Ronald J. Kuivila wrote:
[...]
> > i'm not sure what you mean by sample accuracy in event dispatching;
> > events are quantized to at least the clock resolution, no?
> 
> James explained this a couple of days ago, sample accuracy is attained by
> writing only part of a buffer (that begins at the point that corresponds
> to the timestamp quantized to the sampling period).  OffsetOut does this.
> 
> This can be pretty important for certain kinds of granular synthesis
> effects.
> 
> As far as I understand, it is only relative accuracy that matters: the
> 'first sample' can be anywhere in the buffer but the nth sample must come
> n samples later not n rounded to the control period.  So it suffices to
> take the timestamp, multiply it by the sampling rate, and mod by the
> control period.

ok, thanks. does that mean you're trading sample accuracy for
lateny?  you have to stamp your packets with future time tags
(additional latency worth at least one buffer) to make this
work, right?

<sk>