[LAD] setBfree - Fast Leslie is weird...

[Robin Gareus]

Hi Fons,

I've taken this to LAD, since it's not really a user issue.

On 07/17/2012 12:29 AM, Fons Adriaensen wrote:
> On Mon, Jul 16, 2012 at 10:16:16PM +0000, Fons Adriaensen wrote:
>  
>> It resolves the segfault. Things seems to work, but after a minute
>> or so the sound gets really horrible, as if the whole instrument
>> were detuned. Even octaves sound wrong...
> 
> Seems this was the Fast Leslie, activated accidentally by
> using the sustain pedal. There's definitely something wrong
> with it. The slow version works nicely.

Could it be that you've also accidentally sent other MIDI-CCs to
setBfree that changed the Leslie's behaviour?

The wobble (wrong octaves?) is not unexpected for two independent
doppler-shifts and is also present in the real thing(TM).

Yet, the Lesie is currently indeed the weakest link in the whole
emulation. You're correct in saying that it is wrong. Yet, IMHO it still
does sound OKish.

It's implemented as a combination of angular dependent IR - measurements
of the horn made at CCRMA [1] - which are also [incorrectly] used for
the baffle. Plus the optional static IR to add the [feel of the] cabinet
- which is also wrong, buts sounds good nonetheless :) - I'm hopeful
that we'll get our hands on the IR measurements made for [2] which will
make this 2nd stage obsolete.

The acceleration is also not [yet] correctly implemented. I'm about to
fix this with information provided in [2].

Even though it does sound kind of weird. I prefer it to the mdaLeslie or
similar effects which are far too clean. YMMV.

Your criticism is just and appreciated. Please bear with us - I suppose
that we'll eventually end up rewriting the effects which carry a lot of
legacy code. -- If you can spare the time a DSP expert like yourself
would be very welcome.

Anyway, one step at a time,
robin

[1] https://ccrma.stanford.edu/~jos/doppler/dafx02.pdf
[2] http://www.aes.org/e-lib/browse.cfm?elib=15120