DSP Robotics Support

KG_is_back wrote:

tulamide wrote:Changing the volume knob does nothing

I'm sure I didn't write that!

tulamide wrote:Hey Nowhk,

I'm not sure if all details are clear to you. So pardon my descriptions, if already known:

What you see from the meter is not the actual value from the stream for several reasons.
1) The bi-directional meter is buggy. There's a "minmax" module inside it that maps all values to the 0-1 range of two directions of the meter.
2) A meter always shows as positive values. A signal of -0.5 will be shown as 0.5 just as a signal of +0.5.
3) A meter is just a slow peek at the values. Approx. 25 times per second is the signal evaluated. But a signal has 22050 values per second at least.

A signal consists of values in the range -1..+1, where 0 means no amplitude. At first I thought you'd try to establish some kind of persistent peak level meter (where the last peak is shown until a stronger peak comes in). But you are only interested in positive values in your example. Could you please tell a little bit more? What is the final goal of this test?

Yeah in fact that means nothing. Envelope follower should do the task (because doesn't take care of +/-, right?).

Basically, when I get any signal (i.e. > -INF) I should open a gate, that will do somethings (in my case, trigger a MIDI). And this gate must be opened until I reach again -INF (where I close the gate). And so on.

So, every time I get a signal (from > -INF, or a fixed value, such as >0.5) I trigger a MIDI.
Here for example:


I will trigger 5 MIDI note. Is it more clear what I'm looking for?

tulamide wrote:

KG_is_back wrote:

tulamide wrote:Changing the volume knob does nothing

I'm sure I didn't write that!

Hahah! Yeah of course for my "vision" it does nothing Where am I wrong? In Mono Boolean Module I'm sure...

Yes, that makes it more clear, thanks!

Sadly, I suck at dsp code, but in generic-pseudo-code it would look somewhat like this (assuming a gate at 0.5 and a width of 0.2):



To prevent the gate being switched off right after being switched on, some kind of envelope would also be needed. You can't use any of these informations directly, but maybe the dsp gurus can build something from this?

tulamide wrote:To prevent the gate being switched off right after being switched on, some kind of envelope would also be needed. You can't use any of these informations directly, but maybe the dsp gurus can build something from this?

But this refers to the signal after Envelope follower, right? Envelope follower should just send signal Unipolar, so I don't have -3/+3, -4+4, and so on... so why do I need an envelope? I don't think I need abs?

______________________________

EDIT: Oh no, now I see using the example suggested by KG_is_back. True/False into the Mono Boolean Readout is heavy swapping between True and False with values over 0.5. That's why it process values at different clock, so sometimes is >0.5, sometimes not (since Bipolar, >-0.5).

But that's strange, because neither abs fix the problem; it seems that also with Envelope follower (attack 0, release 50%) sometimes it reads values between the signal. Uhm...

______________________________

I would use a gate of 0.1 (without any width). Once the signal exist (there is sound) this means (for me) that I need to trigger a MIDI. This because I pre-process signal before the Envelope Follower: a bandpass filter + noise gate. So I really get the waveform as I posted above. Signal (from 0.1 or more to 0db or more; else, silence).

It seems I can only make things with DSP code? I believe on using existing modules, but maybe is even better (fast process at low-level code).

I'm a programmer, I think I could do something with DSP code (I'm reading the manual right now).
A thing that I don't get is that "streambool", and how to convert in "just" bool true/false (so I can link to external modules). And also: how can I assign a value to a variable? So at the next clock/stream in/out I can check the state of it. The concepts of "functions" does not exist

Anyone can help me, masters?

At the moment, I'm here:


So, once I go over 0.5 (and index is never used, 0), thanks to pivot I set index to 1. Else it is always 0.
But seems it doesn't works ahahh

KG_is_back wrote:There is something wrong with the formula in the DSPcode part.

Code: Select all

out = (out|(in<0.5))&(in<1);

Unless the initial value is in 0.5-1 range this statement is always true for values below 1 and stays that way until in>1

No, why? The expression is true if in < 0.5, and is false if in >= 1. In between, i.e. 0.5 <= in < 1, out remains unchanged true or false, whatever it was previously. That's called hysteresis, I thought that was the OP's intention.

Guys, no one can help me a little more?

Here you go. I set the volume properties so that the output can now be >1. Also set Attack and Release controls to some value >0 to get a smoother envelope.

I don't know if the OP will be able to work with that test2.fsm, but I see that it is not exactly what was asked for. Currently a gate opens at some value and closes at some other value. But according to the description it should be a gate that opens at a value and closes at the same value some time later.

I hate that I am not able to do this in dsp code, but my pseudo code example (some posts above) would do just that, only that it lacks a timer that makes sure the gate isn't closed right after opening. Is it not possible to do something like my example (plus timing) in dsp code?

Refer to [url]Schmitt trigger[/url].

Tulamide, your pseudocode has two issues: Tt will not trigger if the signal jumps rightaway from, say, 0.43 to 0.5.3. Conversely, if the signal moves slowly through the (0.49,0.51) band, you'll get multiple triggers back and forth.

A code with a timer is of course possible as an alternative to the Schmitt Trigger. This is what I use as a clipping indicator (unoptimized for clarity):