SID Voices

[ralferoo]

Yeah, that sounds awesome. The square waves do make it sound quite harsh, but it's still very nice...

[redbox]

Was this run on a Plus or CPC? It sounds really great! Could you elaborate a bit on how this is achieved technically?  What is this delay thing you mentioned?

This example used the DMA but it would work just the same on a normal CPC if you updated the registers every 64us.  Maybe delay is the wrong word, but I used this because generally by modulating the AY frequency you either delay or shorten the duty cycle of the wave.  I think it's what you referred to it as an offset with your player.

Basically what I did was start with a volume waveform table representing a normal square AY waveform, like this:

Code Select Expand


defb &0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f
defb &0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f
defb &0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f
defb &0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f
defb &0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f
defb &0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f
defb &0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f
defb &0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f,&0f
defb &00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00
defb &00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00
defb &00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00
defb &00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00
defb &00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00
defb &00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00
defb &00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00
defb &00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00,&00


Then I worked out the frequency of the note played by the AY.  For example, if reg 2 and 3 total 80, we have:

Code Select Expand


4000000 (CPC clock frequency) / 64 (line interval) / 80 = 781.25hz


Next I divided this by the AY resolution (15625hz) and the size of the waveform table (256 bytes) to give me the delay (or offset):

Code Select Expand


15625 / 781.25 = 20     256 / 20 = 12.8


Turn 12.8 into a highbyte and lowbyte so you can use it along with fixed point maths to step through the table.

Code Select Expand


highbyte = 12 (&0c)
lowbyte = 256 * 0.8 = 205 (&cd)


If you then stepped through the table at this rate you would produce a tone which is pretty much exactly like the AY tone (i.e. it will be 781.25hz).

However, when I was decoding the ST timers used in SID voices I noticed two things - 1) they always use a fixed 50% square tone (they have to because it's just an interrupt trigger timer) and therefore 2) generally they just use very slight frequency changes which overlap the existing tone to alter the duty cycle.

So, for example the "SID frequency" might be 782.25hz (just 1hz higher).  This would make a highbyte/lowbyte offset for our waveform table of &0c/&d3.  The lowbyte here is increased by 6.  Generally, nice effects are found in the 4 to 8 range (Tao uses this a lot) but this can be altered depending on the frequency of the note being played so you don't suffer cut outs and interference.  The Commando MP3 used a static 6.

Now the AY is generating a tone of 781.25hz and we are generating a tone of 782.25hz.  Play them at the same time (and cut the volume to the AY produced tone) and you have a modulated AY tone aka "SID voice".

[Gryzor]

Wow, Commando sounds really nice! A bit low in the bass region, but a lovely render.

[arnoldemu]

nice rendition of it!

[BSC]

[...]
Now the AY is generating a tone of 781.25hz and we are generating a tone of 782.25hz.  Play them at the same time (and cut the volume to the AY produced tone) and you have a modulated AY tone aka "SID voice".

Have you been doing anything since? Some new demo sounds maybe?

[sigh]

I just listened to Commadno and it sounds really good.


Even more so - it's super useful to hear the comparison between the AY and your SID voices rendering the same tune.

[redbox]

Have you been doing anything since? Some new demo sounds maybe?

Yes, been trying a few things but time a little limited at the moment 

One thing I'm really interested in is the 'interpolation' used in your 3rd demo - could you tell me how you do it (maybe in pseudo-code)...?

I just listened to Commadno and it sounds really good.
Even more so - it's super useful to hear the comparison between the AY and your SID voices rendering the same tune.

Thanks - it's more of a classic 'SID' effect like used on the Atari ST, but built for the CPC rather than just importing the ST routines and timers.  I kind of combined the ST way of doing things with BSC's wavetable approach and that was the result so far.

But using wavetables is the key for the CPC as it makes some really sweet sounds like BSC is demonstrating at the moment.

[BSC]

One thing I'm really interested in is the 'interpolation' used in your 3rd demo - could you tell me how you do it (maybe in pseudo-code)...?

Sure! Here: SID Voices

Joking aside.. The interpolation (of the wave form) is the same in all demos. Just as I described in the post I mentioned.
Maybe you mean something else..? I have the feeling we keep talking at cross-purposes concerning our terminology..

[redbox]

Hmmm, we are doing the same thing

I think your 3rd demo sounds nice because of the frequencies used.

I await the source on GitHub

[BSC]

Hmmm, we are doing the same thing

I think your 3rd demo sounds nice because of the frequencies used.

I await the source on GitHub

Yeah, but we name some things differently I think the 3rd demo sounds nice because it sounds nice ;-) And because it's a sine wave, not square.


GitHub.. what was that again? *scratcheshead* .. Yeah! I know .. I am a slacker in a way ..

[TotO]

Here, how may look a "tracker/demo" on CPC using the CTC-AY... 


[attachimg=1]

[BSC]

How is your new (plus DMA) engine evolving? You haven't posted anything in a while..

[redbox]

How is your new (plus DMA) engine evolving? You haven't posted anything in a while..

It hasn't evolved much further yet I'm afraid.

Too much real life work