Subject: Re: [linux-audio-dev] munlock and mmap
From: Bill Gribble (grib_AT_billgribble.com)
Date: Wed Apr 05 2000 - 18:59:56 EEST
Paul Barton-Davis <pbd_AT_Op.Net> writes:
> If I give up on "audible x-fades", then maybe I can get rid of all the
> hairiness. And you know what ? I checked with some Cubase users, and
> as I would have expected, there is a definite "click" when you go
> punch in/out.
It seems that it would be just incrementally more difficult to do a
"fade out/fade in" instead of either a true crossfade or a click. If
you can fade in/out over just a few milliseconds, you will eliminate
the click but still not have the "sample accurate" monitoring problem.
A dropout of 10 mS or so won't be noticed in a punch-in scenario.
> If you toggle the monitor signal between pre-existing material and the
> recorded stuff, there *has* to be a click, because they are delayed
> w.r.t. to each other, so you have a waveform jump.
Well, the click just comes when you jump discontinuously across a
"large" interval in sample space. Making sure the jump happens at a
zero crossing will do a great job of getting rid of it, and a
fade out/fade in would accomplish that.
It's not too important to me to have a "perfect" sample-locked monitor
stream for punch in, but it would be irritating to have a loud click
which would probably trigger the musician to ask if that was on the
tape, followed by lengthy discussion.
> I haven't yet figured out (or checked) if the Alesis ADATs click, but
> I doubt it.
They don't. I don't know the exact mechanism for avoiding it, but I
doubt it's an analog delay. Probably some trick of fade out/in,
crossfade, or switch-at-zero-crossing.
b.g.
This archive was generated by hypermail 2b28 : Wed Apr 05 2000 - 20:12:28 EEST