� Bla bla bla bla The PCM sample processor, the truth...
����
����
����
����
Those good old days... I can remember them like it was yesterday... Remember those stories in all the magazines when the Turbo-R (ST) was released in Japan? A 28.8Mhz CPU, loads of built-in software and a PCM-Samplechip. WOW!! Now it would be possible to use samples with your music, or voice-recogni sion. Yeah, righty... But still, the PCM isn't at all bad...
Music
Were the music part is concerned, there aren't that many options, unfortunately. Adressing the PCM chip takes too much time for any other things to be done. Music + samples there- fore isn't really an option. It is possible, but the sample freuency has to be very low (No more than 12Khz) and there will be a 60Hz rustle on the back of the sound. For something like a drum, that's not too big a problem. But when it comes to speach or something like that, you'll notice.
The FD#25 game Ryu No Chie is a nice example of what I mean. The sample frequency there was 8Khz, the music was played via a slightly modified Moonblaster replayer in stereo mode. Naturally, that wasn't the best a Turbo-R can do. For one, the R800 wasn't on, and the R800 mode is at least 4 times faster than the Z80 mode. Furthermore, the standard interrupts where activated, which isn't too clever either.
FREQUENCY
The previously mentioned magazines told us the maximum sample frequency on a Turbo-R was about 16KHz. Not at all true... In ideal circumstances a sample (play!) frequency of 400KHz(!!!) should not be a problem. But, like said, under ideal circum- stances...Think OTIR in cases like these. However, under not so ideal circumstances (read, a full-function PCM player) 16 Khz is not the top by far. A sample frequency of 44KHz (CD quality) should not give any problems either. However, the big problem is recording those samples. As the samples are recorded with only one bit, is takes a lot more CPU time to record a sample, than it does to play one... Too bad...
I haven't calculated all this, but I think recording should be possible at a sample-rate of 24Khz, playing has very fiew limits...
HOW TO PLAY A SAMPLE...
Let's make a little program that plays a sample from $4000 till $BFFF. Let's assume we do not need any interrupts, and it is not possible to abort the process. Here's what you'd get:
BEGINN: LD HL,$4000 LD BC,$8000 LD A,%00000010 OUT ($A5),A
MAINLP: LD A,(HL) OUT ($A4),A INC HL DEC BC LD A,B OR C RET Z JP MAINLP
A small player like this one would put you well over the 44Khz C quality. And when I say well over, I mean well over. If I take a short look a this, it would put the sample rate at pro- bably trice the frequence of that of a CD player... But, I could be wrong... (I'm probably not though) Only remaining problem is it won't be possible to record samples on a Turbo-R at that frequence, you'll need quite a PC for that. Other pro- blem is that at a frequence this high, 32kB's of sample isn't really a lot to listen to. But still, that 16KHz really wasn't all that accurate...
Anyway, I suggest that those interested juggle a little with this small source, and we'll continue next time, with a real player with adjustable sample-frequency! How nice...
By the way, that first OUT is to set the PCM chip for:
Play sample, Don't send MIC in to speaker, Don't disable
Sound chips (Music/PSG), activate filter...
Well, good luck... And till next time...
Tobias Keizer