Quote:
Originally Posted by
giulioz
β‘οΈ
Ciao Giulio.
Enjoyed the talk a lot.
You're quite a talented chap.
Tittle is a tad misleading though, since building an affordable (aka no
Kurzweil K250) sample based piano was not trivial in the mid '80s. And the Korg M1 wasn't there yet.
The MKS-20 page at Wikipedia had already some infos from Spectrasonics, but it's been updated as per your findings for MAME, which is great.
https://en.wikipedia.org/wiki/Roland_MKS-20
Quote:
The system effectively implements a wavetable player, able to playback 160 voices at the same time, 10 for each note. For each voice the CPU controls amplitude, pitch and wavetable address.
Every time a note is played, the system reconstructs the piano timbre by layering 10 waveforms played from a ROM chip and modulating their amplitude in time. Some employed waveforms are single-cycle (such as sine waves), while others are short attack sounds. This sound generation method is very similar to additive synthesis.
That's much more than basic sample playback, and offers a lot of digitally controlled expressive means, close to the K250, LA-synth and additive synthesis as you mentioned.
Also, the "voice preserve" aka "patch remain" ability on sustained notes is something that not all current digital synths can do either.
Dave Rossum sure will appreciate those findings and tricks. He always said Roland engineers were very good and creative when facing technical or economic limitations, much like him.
https://www.youtube.com/watch?v=XAuEv29gHo8
Heck, they even had a digitally controlled analog synth that could - almost - do FM like sounds back then (JX-8P) . No wonder hybrid synths are back in the game today.
Quote:
Originally Posted by
OK1
β‘οΈ
But to me, from memory, it sounded like something was missing.
Most likely the analog parts of the signal path (chorus, tremolo and EQ) which are not emulated int the digital domain.