My name is John Sauter. I do sound effects and sound reinforcement
for community theatre groups in New Hampshire, so my interest is in
the semi-pro aspects of the sound card. Almost everything that
spikethehobbotimage has specified meets my needs, so I will restrict
my comments to the few areas in which I disagree with him.
The ability to place breakout boxes up to 100m from the host system
is overkill for me. 10 meters would be plenty.
Consider using AES50 rather than ADAT lightpipe. It is an open
standard, is carried through category 5 cable, includes very good
synchronization capabilities, and holds up to 24 channels, if memory
serves. Google for AES50 for details. It might even be the right
protocol between the card and the breakout box.
The proposal is for the internal format to be 48-bit fixed point, with
a reply indicating that 32-bit fixed point would be sufficient.
I suggest considering 32-bit floating point. Current DSP chips
support it, it consumes no more bits per second than 32-bit fixed
point, and it provides greater dynamic range than 48-bit fixed point.
Using AES50 would solve synchronization issues assuming the card
or the breakout box can accept or provide synch. AES50 includes
both bit synch and word synch.
For my part of the market, MIDI is unnecessary.
If the card implements AES50, then AES50 can be used to link
cards together, either in the same PC or separate PCs. However,
having more than one card is overkill for my part of the market.
In the area of signal handling I regard "live" in my part of the market
to be sending microphone signals to the auditorium speakers, and
"real time" to be sound effects. The "live" specifications are fine, but
the 500 ms delay to "real time" is too high. I currently work with a
300 ms delay, and I need special cooperation from the director so
that there is a 300 ms "warning" for sound effects that must be
matched exactly to the action, such as a gun shot or a paddle smack.
Even with a long wind-up it takes a lot of practice to get the sound
to "hit" correctly. Increasing the delay to 500 ms would make my
bad situation worse. Perhaps there could be a way to pre-load a
sample and place it on hold. When the hold is released the sample
plays with minimal delay. For back-to-back sounds I could pre-load
several effects in a long sample and toggle the hold signal to meter
it out.
If I might suggest a specification for such a pre-load buffer, I think
having five one-second buffers would be sufficient. During an idle
time I could load up to five seconds of sound. When playing, as soon
as a buffer is exhausted it is refilled from the host and goes to the end
of the line. That gives four seconds to refill a one-second buffer,
which should be plenty.
Another application for real-time is recording sound effects in the field.
In this case the delay doesn't matter.
John Sauter ([EMAIL PROTECTED])
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)
