Here's the thing:

Hello,GriV Awesome thing! RULES! I missed it at the time :( What if you make it play from the CD? Then it will be possible to have such a frequency amuse?

Hello Corpsegrinder Doesn't this sound cleaner? The demo was Polish, called MIX #2. File: 1.ZIP http://zx.pk.ru/attachment.php?attachmentid=1909

Hello,GriV Gri> I don’t understand, how about with a CD again? Well, I'm talking about this: Gri> purely theoretically (in fact, it just breaks me to write a program) Gri> you can write a decoder-player that would work with the average Gri> decoding frequency 233.333 Hz = 233 kHz - it's almost like vinyl Gri> (Spectrum is close to vinyl in quality... CD-audio is resting :D Gri> ), accordingly, you can rewrite the encoder in Pascal, but here Gri> here is the calculation - 1) This is a non-packing encoder 2) In connection with this Gri> its operating time for the entire 256kByte of Scorpion memory is something like Gri> 8-9 seconds :(. As far as I understand, the problem is that there is too little memory? What if from CD load all the time?

Hello Surfin' Bird Sur> Awesome thing! RULES! ;) Sur> I missed it at the time:( Sur> Sur> What if you make it play from the CD? Then it will be possible to have one Sur> increase the frequency? I don’t understand, how about with a CD again?

Hello Surfin' Bird Sur> Well, I'm talking about this: Sur> Sur> As far as I understand, the problem is that there is too little memory? What Sur> if you load from CD all the time? If this can be done without the participation of the processor... ;))) then please, how is that? I can't imagine how to do it without the help of the proc ;))) If anyone has launched it on a real Speck, they will notice that there is a third of the disk there at the beginning into memory - this is almost 30 seconds - then playback just starts - the same 30 seconds in the first sample and a minute in the second... This is exactly what won’t be able to load; for video, the discreteness when loading is not the same very noticeable, but unfortunately for audio...

Hello Surfin' Bird Sur> Well, I'm talking about this: Sur> Sur> As far as I understand, the problem is that there is too little memory? What Sur> if you load from CD all the time? There is also an option that there is some data that is fed to the port - so that they can be taken from there without management (at least permanently) remove (with a simple command IN A, (port)), this is also possible - it will even be easier - because it’s more convenient to work with this later - you don’t have to monitor memory limits - and the program replies to poll the keyboard at interruptions - then it’s totally cool. But I doubt it will be that simple...

Hello,GriV GriV, you're great! :)

Hello, CityAceE Back in Born Dead 2 [http://trd.speccy.cz/press/borndead/BORN_02.ZIP] there was about this, with a demonstration.

Hello,GriV P.P.S. I don't want to get ahead of myself... but I'm almost finished even more a productive algorithm, although I haven’t made samples for it yet - I collected decoder only. When I assemble the encoder and debug it, I’ll post it... by all standards, this is it will just sound great... I had more in development productive in terms of speed (almost 1.5 times faster!) circuit, similar to the one given two posts above, but 4-5 seconds of sound (at 256 kb!) I think few people will be satisfied - it’s simply impossible to distinguish the sound level from the previous version - that's why I didn't even develop it.

Hello,GriV Respect. sane. but you can try to make loading from the HDD in real time - it won’t be much different from switching banks (Anril, by the way, knows how to cheat HDD). and if perverted, then simultaneously with loading from the flop! (option - cache+turbomod)

Hello, The Exploited no, the steering wheel is unquestionable. Especially when the memory is 1m. And if the speed of working with hard drive is really non-lethal for sound, then the result could have been devastating. But here you probably still need DMA...

Hello TomCaT don't swing

Hello, The Exploited Yeah, they don't swing. Check it out.

Hello, The Exploited yeah, the links are dead... :(

Hello Savage It's my fault, but not mine. My domain name is fooling me - it seems like the registration has passed, mail is running (MX record works) but for some reason NS does not resolve. Replace the given links instead http://say.da.ru/ to http://grivalera.da.ru/ P.S. It seems like I fixed it this way and that, but it’s not clear why it doesn’t work. With the same settings grivalera.da.ru works... P.P.S. A day after this message, everything should work according to the above addresses, it seems I found the reason why the links are glitchy...

Hello Orionsoft Now I have one idea in the air... rather like delirium. The meaning is this - the most "moronic" command in all coders is the transition to next output cycle. Those. Or it looks like this: Out(c),X Ret - as many as ten clock cycles with 12 outputs or Out(c),X .. - here are some calculations for the transition, which by definition is greater or equals 4 measures Jp (HL) - 4 more bars, total 8 So I thought for a long time how to get rid of it - and came to the only conclusion - it’s simple remove Ret. :-D Now the sound output goes without Ret! And for this, the encoder program generates a ready-made Z80 code - i.e. Out commands (c),X NOP and others for stuffing with clocks for clock intervals: Out (c),d ; speaker rise nop ; delay 4 ticks until it reaches the desired position Out(c),c ; lowering Inc hl ; pause 6 clock cycles, hl is not used at all - only registers C and D are popular Out (c),d ; here, immediately after the rise, there should be a descent without a pause Out(c),c ; it is Nop nop ; pause 8 bars for lowering Out (c),d ; another rise etc. - this is exactly what the encoder program will generate The only thing holding me back is that this code will be terribly suboptimal - about 2-3 seconds of playback for the whole body in memory of about 1 meter - but it will be as accurate as possible. In addition, unlike previous versions, it hasthe ability (at least theoretically) to unpack after downloading, i.e. in fact, a full disc will take about 20-30 seconds of sound, after each shipment will be unpacking into memory and playing itself. Yeah, I almost forgot The trick is that almost all previous (except Baba Yaga's song) samples were in constant loop size, no. Here the cycle size is not constant, but very That is the discreteness is as follows: The first output to the port, so of course the length with the duration of the cycle size equal to 1 equals 12 clock cycles - Out (c),X Cycle duration for 2 - 16 bars - Out (c),X and Nop and then there is an addition under two measures - 3 size - 18 measures 4 - 20 5 - 22 6 - 24, etc., and since ready-made code is generated, there are restrictions on there is no cycle size - at least 493859836986th size ((((((-; Huge gaps (they possible only theoretically, in practice they simply do not occur) it is possible even code with loop commands like ld b,12 djnz $

Hello,GriV Not everyone's M1 clocks are rounded in real life...