Sound channel modulation: Ever used?

[NorthWay]

Does anyone know of any demo or other program that actually used the sound modulation abilities?
Thanks.

[Toni Wilen]

Never seen or heard it being used. Only existing program I know was test program to confirm and fix emulation

[Lonewolf10]

Just curious, but is what is sound channel modulation?

[Tumo007]

Music from TurboImploder 4.0? I think Paul van der Valk used it in is custom player code. I might be mistaken...

[Michael Sykes]

Originally Posted by Lonewolf10

Just curious, but is what is sound channel modulation?

Yup... that would be interesting to read.
If anyone could enlighten us ... please!

[Don_Adan]

Originally Posted by NorthWay

Does anyone know of any demo or other program that actually used the sound modulation abilities?
Thanks.

SoundPlayer by Scott Johnston used sound modulation code.
But I don't know if any SoundPlayer module used this feature.

[NorthWay]

Originally Posted by Lonewolf10

Just curious, but is what is sound channel modulation?

Well, if Toni has implemented it he can probably give a better/correct explanation, but it is basically using data from one sound channel to shape the data of another sound channel.

As the RKM tells it; you have a choice of period or volume modulation, or both taking turns.
My interpretation of that is that you can have a very "funky" playback of a sample by having the frequency of it jumping all over the place (though I'd think it would sound better if it was following a line/curve), or you can have the volume of a sample be changed very dynamically to achieve some effect I can't think of right now (not echo like?). A smooth fade in/out possibly.

The question that begets is if the modulating sample is played or if it becomes silent? You might possibly achieve "ring modulation" like effects.
This all sounds like clever ideas someone once had, but are they really useful for sample playback where you can just shape the sample yourself? It sounds like a design for some _very_ small memory, like the original 128K plans, where I could envision the idea was to spend no more than 8-16K for all sound (code included).

And does this achieve something, running at speeds of once per scanline, that you would not be happy with doing just once per frame?
A chiptracker could probably find some use for this, but I don't know of any that used it. Hence the Q.
(Would it be possible to have UAE show some statistics which could reveal any such use?)

[Toni Wilen]

Modulator channel's DA converter is always disabled. Which probably explains why it wasn't used, number of channels gets practically halved.

[pandy71]

With correct data alignment and accepting fact that sample rate is halved this mode can be used to create fully hardware 14 bit.
(i miss channel combining function - two DMA feed data in DA - or for twice higher sample rate or for 16bit - perhaps in new Paula this can be added)

[NorthWay]

I was about to write something or other, but then I got to wonder why it is called 14-bit mode? Isn't it just the upper 8 bits played at full volume on one channel, and the lower 8 bits played at a volume that is the dB equivalent of 1/256th? Isn't the channel volume mixing linear? Isn't a volume of 1 reduced dB-wise to close enough to 1/256th?

Still looking for a modulation demo though!

[Lonewolf10]

Originally Posted by NorthWay

As the RKM tells it; you have a choice of period or volume modulation, or both taking turns.
My interpretation of that is that you can have a very "funky" playback of a sample by having the frequency of it jumping all over the place (though I'd think it would sound better if it was following a line/curve), or you can have the volume of a sample be changed very dynamically to achieve some effect I can't think of right now (not echo like?). A smooth fade in/out possibly.

Thanks. Looks like I'll be reading the RKM this weekend

[pandy71]

Originally Posted by NorthWay

I was about to write something or other, but then I got to wonder why it is called 14-bit mode? Isn't it just the upper 8 bits played at full volume on one channel, and the lower 8 bits played at a volume that is the dB equivalent of 1/256th? Isn't the channel volume mixing linear? Isn't a volume of 1 reduced dB-wise to close enough to 1/256th?

http://amigadev.elowar.com/read/ADCD.../node00F1.html

http://amigadev.elowar.com/read/ADCD.../node00DC.html

http://amigadev.elowar.com/read/ADCD.../node0017.html

PWM level modulator is only 6 bit and only those 6 from 8 is used.

http://amigadev.elowar.com/read/ADCD.../node0012.html
http://amigadev.elowar.com/read/ADCD.../node00E7.html

[Mark Wright]

As Tumo007 has said (and I expect Don can elaborate if he feels the need..) I always assumed Paul van der Valk used this technique to create his signature "space" sound, common to his later works. It was obvious just by listening that he was somehow exceeding the four channel limit without the loss of fidelity associated with the tell-tale scratchy sound of the Oktalyser mixing method. Turbo Imploder even had a visual analyser showing the five channels as the music played.

It's probably documented elsewhere by now, but I can remember my admiration for whatever alchemy went into creating the Turrican 2 theme when I first heard it. It took me a while before the penny dropped that the unusually spacious sound was due to straying beyond the four channel limit. It had none of that give-away, lo-fi scratchiness, and there were other things happening on-screen as the theme played. Was it the magic of TFMX, or just careful planning on the part of Chris Huelsbeck - or a mixture of both?

Sorry, I realise the above is off-topic, but what a shame that more classic Amiga songs using the "acceptable fidelity" seven voice exploit weren't made. Thinking back to when coders discovered methods of replaying samples on the SID and AY chips - together with Tim Follin's awesome 1-bit workouts on the ZX Spectrum beeper - I wonder, did coders optimise the abilities of a stock A500 Paula to within an inch of its life back then, or are there still more theoretical improvements yet to be made.