Blitter sound routine

[metalages]

I shared some ideas here about using blitter for 2 pcm channels sound mixing here :
http://cyber.savina.net/sound.htm

... and also some proto code here :
https://github.com/gibs75/demOS

What do you think ?

[AtariZoll]

Interesting idea. But I don't think that on STE we have much possibilities, because it is slow for any serious processing of audio data.
As title says "mixing" it reminds me that I spent some time for code what mixing background music and sound effects. Would be good if blitter could make it faster.

[dml]

I shared some ideas here about using blitter for 2 pcm channels sound mixing here :
http://cyber.savina.net/sound.htm

... and also some proto code here :
https://github.com/gibs75/demOS

What do you think ?

Interesting. I also found it wasn't much use for adding - but if you move the aim from real-world samples towards YM-like synthesis from short waveforms then you can do a few interesting things. e.g. using smudge + halftone to perform modulation effects on 4-bit 'samples' - or perhaps more interestingly - on the most significant bits of 16-bit samples!

I've mostly been working with 8->16bit samples on F030 so it's not that much use to me in this form but I think there is room to explore more here. Good stuff. Keep investigating

[FedePede04]

what a great idea, keep it up the good work

[metalages]

Interesting. I also found it wasn't much use for adding - but if you move the aim from real-world samples towards YM-like synthesis from short waveforms then you can do a few interesting things. e.g. using smudge + halftone to perform modulation effects on 4-bit 'samples' - or perhaps more interestingly - on the most significant bits of 16-bit samples!

Yep. Using small synthetized samples are possible as the blitter loops for free ; also it would prevent from excessive memory usage due to the need of pre-transposing samples for semi-tones.

[metalages]

Interesting. I also found it wasn't much use for adding - but if you move the aim from real-world samples towards YM-like synthesis from short waveforms then you can do a few interesting things. e.g. using smudge + halftone to perform modulation effects on 4-bit 'samples' - or perhaps more interestingly - on the most significant bits of 16-bit samples!

I've mostly been working with 8->16bit samples on F030 so it's not that much use to me in this form but I think there is room to explore more here. Good stuff. Keep investigating

Just to be sure I well understood your post DML

You mean using smudge mode with HOP = 3 to combine two waves with an AND operation :
- the source wave (8 bits resolution)
- a second wave with 4 bits resolution that address value set in the halftone

Am I right ?

[dml]

Sort of - the smudge provides a 4-bit indexing operation on the source, so you can fill the 16 words of halftone with any translation (e.g. invert 0->15, punch holes in the table at certain volumes, or make a saw/ramp with it) and then use scroll to access the bits in the sample you want to translate, and endmask to replace just the bits of interest. I was thinking about pointing source/dest at the same sample, but separating them would let you combine two samples in a limited way.

But it seems that you can do a variety of even stranger things with this. Since the halftone table is indexed with the y (line) field, you could use x:y inc ratios to scroll the translation - which gives you another (sortof) waveform period for the translation. A bit like what you describe on your page re: periods for short samples except using indirection via halftone ops.

Of course none of this maps to typical audio mixing/processing steps and it's probably hard to find combinations that make useful sounds. Still it looks fun to explore.

[Mikefulton]

Despite the limitations I have to say it's a very clever idea!

[metalages]

I have updated the page here http://cyber.savina.net/sound.htm and have a link to the PC executable version of the prototype (at the bottom of the page). In case you want to give it a try...
Same code takes an age to generate samples on STe (even under emulators at full speed) for the moment but then it mainly works the same way.
Of course currently these samples are not interesting in musical terms
The channel 2 features the dynamic use of halftone to mask less significant bits.
Currently it does not demonstrate the use of bitshift to change volume (but this can be used in the prototype)

[metalages]

New version of the prototype : to allow experimentations by musicians without coding, the new prototype use a script files + samples instead of generated sounds.
Documentation into the script (SYNTH.INI)

Links here: http://cyber.savina.net/sound.htm
PC executable here: http://cyber.savina.net/synthsound/synthsound_proto.zip

[metalages]

The prototype now runs on STe (tested on my 2mb STe) and PC. New version submitted (sources are on github).

STe + PC executables are here: http://cyber.savina.net/synthsound/synthsound_proto.zip
Links here: http://cyber.savina.net/sound.htm

This prototype is a proof of concept => its allows to experiment with your own samples to see if something interesting is possible with this approach.
Just edit SYNTH.INI script file and add / modify your samples.
Current script + samples are designed to be similar with my first prototype (1st proto was working with generated sounds).

Currently DMA runs at 50khz. It is possible to save CPU and memory using 25khz...

[metalages]

I have just added some rasterization on STe here http://cyber.savina.net/sound.htm
(playing at 25khz and 50khz)

[metalages]

And the optional effects you have for free from the same sample are :

- decimate bits with half pattern mask
- shift bit (reduce volume)
- octave transpose

[metalages]

... And of course managing small samples efficiently by looping for free.

[MM41]

Tested on my STe 4mb, nice work Metalages

[metalages]

Thank you
It is a proof of concept more than a tool for the moment.
Anyway the new version with script and samples allows some user experimentations

[metalages]

Dok told me to look at wavetables synth to get some ideas.
For example: https://www.youtube.com/watch?v=R2Ka3wJgwIk

As you can change the sustain sample on the fly without restarting the sound (W => ; keys on AZERTY keyboard)
it would be possible to create a set of pre-interpolated waves.

In the provided test, this is the case for the first sound (3 samples can be choosen as sustain on the fly)

[metalages]

I have rebooted my experiments on a more traditional approach using standard protracker .mod format as input.
https://youtu.be/HgQVavtUOGI

[Cyprian]

sounds cool, especially distortion

[ggn]

This is absolutely great! Great work metalages, and IMO much more interesting than the usual what-happens-when-you-flip-this-shifter-bit-at-exactly-29.3-cycles-from-the-vbl-and-it-totally-works-in-my-machine-yours-is-bad-boooo kind of research that people usually do with the ST/E .

Also, I do have some projects in the pipeline that could benefit from such a sound routine so thanks for sharing this

[shoggoth]

Wow!! Amazing stuff, new ground. Hats off!

[metalages]

Thanks for your kind words

I have huge respect for the reverse engineering work some people do, but it is not my stuff indeed
I prefer trying new stuffs. Even on a small machine like the STe some stuffs have not been tried yet.

[metalages]

This is absolutely great! Great work metalages, and IMO much more interesting than the usual what-happens-when-you-flip-this-shifter-bit-at-exactly-29.3-cycles-from-the-vbl-and-it-totally-works-in-my-machine-yours-is-bad-boooo kind of research that people usually do with the ST/E .

Also, I do have some projects in the pipeline that could benefit from such a sound routine so thanks for sharing this

My idea was trying to do a screen with this routine and then putting the whole code on my github depot.
Let's keep in touch about that

[Maartau]

Really great .

Hats off [smilie=greencolorz4_pdt_01.gif] .

[charles]

any chance youll adapt midi as the file format ?
have loaded spl/wav triggered via midi notes ..?
I guess like gadjits seq one and other commercial tracker/sequeners for Atari?