[sdiy] In praise of the ATM STM32F303

[Michael Zacherl]

On 12.Feb 2016, at 19:06 , rburnett at richieburnett.co.uk wrote:

> Just watched the Streichfett teardown video, and it is very impressive what they've managed to squeeze out of that processor!!!
> 
> Judging from some of the comments regarding aliasing on Muff Wiggler forum, I'd guess that it just uses naive software generated sawtooths without any particular measures taken to attenuate aliasing.  Or possibly just a tracking HPF filter to clean up the spectrum below the fundamental frequency like Adam Szabo found in his JP-8000 Supersaw analysis.  This is quite cheap to implement, particularly if you've got double-precision (32-bit) processing at hand...
> 
> "I recently got a Streichfett, and I can also definitely hear a lot of nasty digital aliasing type sounds from the String section. There are a few sweet spots, but it still takes a decent EQ to reduce some of the unwanted sounds.
> The Streichfett doesn't compare to any of the vintage string synths
> I've played (Rolands and Elkas), and unless Waldorf updates the OS to clean up the aliasing I'll probably return it.
> The Streichfett lacks a low-pass filter and it sounds horrible in the highs (aliasing)."

http://www.soundonsound.com/sos/dec14/articles/waldorf-steichfett.htm :

"If there’s a limitation, it’s the low bandwidth of the instrument. Select one of the brighter waves and play up the keyboard, and you’ll find that the level of the sound soon diminishes. What’s more, the amplitude of the notes at higher pitches can be very uneven. I was using a vaguely choral waveform somewhere near the one o’clock position of the Registration knob, and I noticed that the output of an upper ‘E’ was many, many times louder than that of adjacent notes. The difference was so marked that I measured it. First, I played an A: its output was 138mV. I then played the C three semitones above it, and the output was 285mV. I then played the E above that, and the output was 1.05V. Consequently, the upper note of the triad dominated the A/C/E chord, which is quite unlike the response of a traditional string synth. This appears to be a consequence of the formant shaping for the vocal waves and, if so, is understandable, but you should nonetheless be aware of it.”


--
http://mz.klingt.org