[sdiy] Pink Tunes self-composing program.

[rsdio at audiobanshee.com]

I think it’s worth pointing out that Pink Tunes did not run on an OS. It’s what we all “bare metal” code. There was a common library (MUS-1) used by multiple programs that were peers of Pink Tunes, but that was more of a hardware driver than an operating system.

All of this talk of using an expensive, power-hungry FPGA chip, Java code, or MAME, to emulate the 6503 is really missing the point. The bulk of the Pink Tunes / MUS-1 system was written to interface to custom hardware with a DAC that only supported 64 discrete CV output levels, and it doesn’t seem like much sense to bother with emulation just to interface to such limited capabilities by today’s standards. Not to mention the fact that a perfect 6503 emulation still would not produce any output without matching emulation of the DAC hardware.

Another big chunk of this old code is the software envelope generators. While they have some unique features that aren’t available on analog envelope generators, I somehow doubt that the distinctive sound of Pink Tunes is highly dependent on these specific digital envelopes. I assume that any standard MIDI synth could be programmed with patches that would sound good with the Pink Tune random note generator.

In my estimation, if you boil down the essentials of Pink Tunes, only a small part of it is the 6503 code or the 64-level log DAC. The bulk of what make Pink Tunes interesting is the choice of notes and the way that the random number generator works to build chords and melodies on the fly. Any decent programmer should be able to recreate that using MIDI output, and just skip all of the painful details of the esoteric hardware or emulation.

The text for the Pink Tunes article even mentions ways to alter the code to change the way the randomness is implemented. These sorts of details are way more important than emulating the 6503 QuASH hardware. See the “Voss pink-ing algorithm."

A decent programmer should be able to recreate Pink Tunes in Max/MSP or even a bog-standard MIDI program.

Brian Willoughby

p.s. I think I realize why Pink Tunes only uses 4 voices. It probably has nothing to do with the Prophet 5 voice architecture (5 nearly-identical voices with 1 attached to external CV inputs). It seems that the QuASH circuit in the hardware platform handles multiples of 4 outputs. Typically, a pair of outputs is used for a voice, so Pink Tunes is clearly using 2 QuASH circuits for 4 voices. I assume they could have made it support 5 voices by building a 6-voice system and just turning off 1 set of voice outputs, but that was probably an “expensive” experiment in that decade. The code does seem to support intelligent voice allocation.


On Jan 8, 2018, at 10:43 AM, David Bulog <d2ba at xtra.co.nz> wrote:
> So can Pink Tunes be reborn to run on a modern day OS ?
> 
> On 9/01/2018, at 1:28 am, Tom Wiltshire <tom at electricdruid.net> wrote:
>> I wondered about this. Assuming a 6502 running at 2MHz, if I used a modern PIC with an instruction rate at 8MHz, if I could write equivalent macros for each of the 6502 instructions in 4 instructions or less, I could basically compile the old code for the new chip.
>> 
>> It’s almost (but not quite) an interesting enough challenge to warrant giving it a try. Maybe if I had more Sundays in the week.
>> 
>> Tom