on board sounds

Dimitree · Joined: Feb 09, 2011 Posts: 8 Location: Italy

hi everyone
this is my first topic here and I hope I haven't used the wrong section!
I'd like to know how it work a sound module like those used on board of some keyboards (for example M audio KeyStudio 49i). Is that a microcontroller loading recorded samples from a ROM, and playing them with a DAC? is every note recorded, or are them shifted? how about velocity and sustain? it's a compromise of quality-storage (more storege available, more samples, less shift, and so on?)
many thanks everyone
Dimitri

Dan Lavin · Posted: Thu Feb 10, 2011 9:56 am Post subject:

Welcome to the E-M Forum!

yes, that's basically how they work. They generally have a limited amount of samples and do some stretching to cover in between notes. Velocity can be covered by different samples or an algorithm to change harmonics. If you google "rompler" you'll probably get a better explanation. This technology is 20 years old now (Emu Proteus and soundblasters)

Dimitree · Joined: Feb 09, 2011 Posts: 8 Location: Italy

many thanks Smile

is the stretching performed with DSP hardware? or there are other simpler technics? is this different than wavetable synthesis?

Dan Lavin · Posted: Thu Feb 10, 2011 2:06 pm Post subject:

Not sure how they stretch, but wavetable synthesis is what it is.

Dimitree · Joined: Feb 09, 2011 Posts: 8 Location: Italy

is this something DIY-able (maybe a DSPic with SD card support and good DAC)? I know buying a EMU rack would be cheaper and easier, but the fun would disappear Very Happy

Dan Lavin · Posted: Fri Feb 11, 2011 2:00 pm Post subject:

Dimitree:

We already did this back in 2007!:

http://electro-music.com/forum/topic-21634.html

Dimitree · Joined: Feb 09, 2011 Posts: 8 Location: Italy

many thanks! looks great, I will study it to understand if it fits my needs!
I was wondering, digital pianos like clavinova work like a pure rompler?
how poliphony is achieved in software? mixing the waves of 2 or more files?
I saw that software samplers like Kontakt or Sampletank use 1 sample for each pair of notes (more or less), so there's a minimum amount of stretching, while, if I understood good wavetable synthesis uses just 1 waveform and stretch it for every note. is it right?

egasimus · Joined: Feb 11, 2011 Posts: 106 Location: Bulgaria

the very "best" way, if you're going in that direction is to have one or even multiple samples for each note. but that's sampling. wavetable synthesis is a compromise, and the "best" wavetable synthesis is the best compromise you can get between availabe memory, processing power, sample rate, etc. you could use a wavetable synthesizer instead of a VCO - that would let you use non-standard waveforms.

http://elm-chan.org/works/mxb/report.html

check this out, too. built one a year ago, works great.

28612_1330960999274_1388488725_30896829_2458666_n.jpg

Description:

Custom-built wavetable music box. Behind the mask there's a perfboard with an ATtiny85, a 2032 coin cell, and little else. Note the ingenious amplification solution (metal bell on top of the speaker).

Filesize:

110.66 KB

Viewed:

69 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.