Re: rSounds - affect on available free memory

[johnlb@primenet.com (John Bowling)]

In article <29MAY199606203391@vax2.concordia.ca>,
Mitchell Spector <spec@vax2.concordia.ca> wrote:
>	Ever since System 6.0, the 'Sound' CDev/Control Panel has 
>allowed users to assign digital sound effects to system events 
>(using rSounds). All very nice, but I noticed that once a sound
>is assigned and loaded into memory, that memory becomes reserved
>and leaves you with less free memory for applications. Or at least
>the Finder reports this when selecting "About this Apple IIgs". 
>I also noticed that any sound assigned to 'Startup Sound' (which 
>is used _once_ when starting GS/OS) is taken and never given back.
>If your startup sound is 2MB in size, you'll have 2MB less available
>free RAM. 
>
>	Are there any patches to change this? Perhaps something which
>would have GS/OS only load sounds _when_ an event occurs and then purge 
>it from memory. I would think my hardrive and RamFAST would be quick
>enough where there wouldn't be much of a noticable delay. In the 
>meantime, I've avoided the use of rSounds for system events (expect
>one: a small sound for the system-beep) because it quickly eats _much_
>of my available memory and even adds a few seconds to my boot time.
>I can understand using up memory on DAs, INITs and Finder-Extensions,
>but for some cute sound effects...? If I'm not mistaken, even Mac
>System 7 and Windoze 95 (which have a similar feature) do not lose
>any or much RAM for system event sounds.

Mac and IBM systems may keep the data compressed in memory.  They also 
use a larger wave memory for sound any may leave some sounds permanantly 
installed there.

The standard system-beep is not a 'sound' like the others, but a software
manipulation of the speaker as it was done in other A2s.  It doesn't occupy
any additional memory if it's selected.  There is a possibility that other
sounds can be created in this way, being more efficient with memory by 
software manipulation.  As an example, a 'whoosh' could be done by taking 
a small clip (1 cycle) of sampled sound, and storing it into the Ensonic 
RAM repeatly with slight mods for each additional cycle (level, duration, 
etc.).  This would take less memory, but would require time to create the 
sound.  If small enough, several sounds could be kept in the wave ram 
during normal finder operation and reloaded on quitting back to the finder.

Another trick could be to slow down play back and shorten the wave by 
removing samples.  I'm not sure how far you could go with this until it 
became obvious, but I would suspect that all of Apple's supplied sounds 
could be reduced by at least half their size this way.

John

--
--==--==--==--==--==--==--==--==--==--==--==--==--==--==--==--==--==--
- John Bowling          -- Nothing worthwhile came to mind, so this
- johnlb@primenet.com   -- is a big blank.  Ignore it!
--==--==--==--==--==--==--==--==--==--==--==--==--==--==--==--==--==--