Subject: Re: [linux-audio-dev] file sizes and retrofits
From: Juhana Sadeharju (kouhia_AT_nic.funet.fi)
Date: ma helmi 07 2000 - 07:35:21 EST
If somebody want to look at large file handling, then I suggest
following approach or something close:
Basic idea is to access audiofiles through our own system which
works the same way as virtual memory systems.
There are a plenty of choises for managing buffers:
-one buffer pool which is always used at page faults;
-one buffer pool from which sub buffer pools for individual
files are made; sub buffer pools are used at page faults, and
the master buffer pool is used only when sub buffer pools are created;
-sub buffer pools for each file instantation; if you play the same file
on several tracks with different offsets, it is better to have
own sub buffer pools for each of them;
-buffers can map randomly to files, or they can map only to locations
where location == 0 (mod bufsize)
-sub buffers can point to random locations, or they can point to
consecutive buffers (mod bufsize); this is not far from a ring buffer;
etc.
Audio can be prefetched or kept in memory prior playing:
-start of the files and loops;
-if you need to tricker the same file multiple times for simultaneous
playing, you keep the start only once in memory and have multiple
ring buffers for playing the rest of the file.
There are many combinations, of course.
The Gigasampler patent seems to have something similar but you can safely
use any of the above because I formulated them mostly at 1990-1991.
The realtime playing part was formulated as above for sure before 1995
because I tested the realtime playing by writing a realtime synth at that
time. However, this synth didn't implement the whistles because due lack in
programming skills we had to restrict to keep whole samples in memory.
(The synth just reminds me what was figured out before that time.)
Gigasampler patent is from 1997, two years later.
Well, patents are business. They seem to advance the fact that non-patented
invents are not documented well or are not spread widely enough.
Another patent which buggers me is a recent patent which decreases the
transmission bandwidth of digital movies by compressing more on areas
where audience/viewers doesn't look at. Only center of action where audience
is supposed to look at is transmitted more properly. This idea was
formulated by me and some other guy at early '90.
Juhana
This archive was generated by hypermail 2b28 : pe maalis 10 2000 - 07:23:27 EST