Subject: Re: [linux-audio-dev] extending LADSPA, it's not that easy ......
From: Paul Barton-Davis (pbd_AT_Op.Net)
Date: Sat Oct 28 2000 - 03:45:10 EEST
>I am talking about something much more basic, the ability to select
>regions for any purpose with single sample accuracy
>within the context of a multitrack environment that advances audio in
>frames larger than one sample, which I will refer to as a
>superframe to distinguish it from a single sample frame.
Ardour delivers data to/from the audio interface in chunks of
"frames_per_cycle", and delivers (most) data to/from the disk in
chunks of "minimum_disk_io_bytes", but all editing operations are
completely sample accurate.
>So are you saying that an edl enabled multitrack host can resize the
>superframe immediately before playback of a new region so that the start
>point of the new region falls exactly at the first sample of the next
>superframe? My mental model has been of a constant superframe size
>based on the fragment size.
The editing operations can do anything with individual sample
resolution. They can insert individual samples, zero individual
samples, etc. etc.
When playback occurs, the EDL-read operation gathers the samples from
the various locations where they are held, and hands them to the audio
engine as a continuous stream. And because this is happening in a
different thread, it doesn't matter much if it takes a little while.
As long as we have a comfortable time margin for the disk i/o, then
spending a little while converting the EDL into a sample stream while
the audio thread is asleep is just fine, even on a UP system.
Its pretty simple, really :)
And its fun trying to find precise sample loop points by doing
1-sample increments and/or zeroing individual samples :) I'm just
kidding, its no fun at all!
--p
This archive was generated by hypermail 2b28 : Sat Oct 28 2000 - 04:34:45 EEST