Generator of background noise (Noise spammer)

Джим · Post by **Джим** » Thu Oct 01, 2009 20:33

In principle, you can take-based timeline with sofronitskij and binaural tones in a textual form similar to SBaGen (named blocks)...

Note that scripts SBG, with all the advantages have not quite obvious, but lethal the disadvantage-there is the whole way it is necessary to calculate the time and carefully fold hh:mm:ss-it is wildly lacking, and it should not happen again.

...Without cubic interpolatio values of the main signal (noise) there is not enough. ...

I never heard about the fact that signal processing in General (and sound in particular) using a standard mathematical interpolation (except when zooming). Ie, any time there is a conversion is always performed in my lifetime in terms of filter sin(x)/x. Maybe I'm behind the times -- a reference would be thrown in handler, where you would use the cubic (or other) interpolation? (I write without irony.)
Guest

And how do you like "FractMus 2000" and "Fractal Tune Smithy"?
There is another quite interesting "musinum208"...

Thank you! But for today I like the prospect of writing it yourself. Ie, learn, learn and learn. Perhaps, of course, I would be tempted ready source of one of these prog, but. Thanks again.

Волутар · Post by **Волутар** » Thu Oct 01, 2009 21:13

How else can you delay the signal in time by 0.3 sample? The easiest and fastest choice is to use linear interpolation, but it is fraught with some artifacts. Usually do cubic interpolation. All sites dedicated DSP is done in this way. At the same musicdsp.org.
When you create the Flanger effect (no matter feedforward or feedback) is taken the same cubic interpolation.

About terms of filter sin(x)/x don't know.

there all the way, it is necessary to calculate the time and carefully fold hh:mm:ss-it is wildly lacking, and it should not happen again.

I'm not very deep poked the convenience of their scripts. What is drive and why carefully fold?

Джим · Post by **Джим** » Fri Oct 02, 2009 0:02

How else can you delay the signal in time by 0.3 sample?

A situation when the required delay to the fractional number of counts, need, and usually can be avoided at all costs.
Code reviewed, it's funny. When strangle all (this is a joke

) -- write out and see -- at all -- wild and new. (And you can say, without hesitation -- what would such a transformation atch? Don't know, but strongly suspect that quite unexpected, with high probability.)
Anyway -- thanks for the response.
naschet sin(x)/x -- this is the ideal interpolating filter (more precisely-his imp. har.). I've never done peridiscaceae, but I know at cu. 2 example, when Norodom proud to say that that is what they used. A specific example is in the code SBG -- where the waveform points are specified-not the best, by the way-there is a window missing.

What is drive and why carefully fold?

If You write a session to SBG containing either loops or just repetitive -- You, of course, there is a reason to use "structure" (named sequence -- what is written in curly brackets). When they are in the main sequence, or when they invested -- time accounting becomes very unpleasant. For example see, for example, 'prog-chakras-1.sbg', but given that there are times artificially aligned. In real life, if you want to give each segment a random time... the horror.

Джим · Post by **Джим** » Fri Oct 02, 2009 0:28

Valutar
Actually, I just remembered that this Forum is dedicated to support all the different devices and the specific CD. Our conversation, therefore, in the framework of the Forum, it becomes almost indecent -- people tend to voprinimat such as a show off, and they are usually right. (Yes, probably my fault, but it does not change).
Personally, I raped it a good idea to paddle towards the artificial harmonic music, for which, one way or another, thank you.
If there is anything interesting -- will manifest. And in the most extreme case -- I have a PM (I'm not DSP guru, but this year all the way have to remember what you've learned and learn a lot beyond that

- not to work, and on the topic of the forum -- on prah hobby type).
Ie, good luck.

Волутар · Post by **Волутар** » Fri Oct 02, 2009 6:42

need, and usually can be avoided at all costs.

Impossible. 0.65 MS is a maximum of 28 samples. Discreteness for HRIR delay is excessive. The desired proportion of such samples to the 3D positioning was more or less accurate. I'm sorry that You are not very familiar with DSP transformations, and "delays" (this is not resampleimage, not the spline interpolation, and even not filtration at all, so sin(x)/x here out of place)

what would such a transformation atch?

Must remain original. Frequency does not change. Changing the delay. Flanger (or comb filter) is created when combining a signal with a slightly delayed copy of itself (with some backward and forward linkage). In this case, the frequency response of this combination looks like "sawtooth". Predictably, the sound and General effect.

For example see, for example, 'prog-chakras-1.sbg', but given that there are times artificially aligned.

Thank you for the info. Be sure to study this item.

The scope of the forum.. I thought this subforum is dedicated to different programs. I have this "different program", which is only at the initial stage and therefore does not yet have binaurally.

Соловей · Post by **Соловей** » Fri Oct 02, 2009 10:26

Джим wrote:Valutar
Actually, I just remembered that this Forum is dedicated to support all the different devices and the specific CD. Our conversation, therefore, in the framework of the Forum, it becomes almost indecent -- ...

Not so much decent that stops on a half-word is still more than decent. Please, continue, here, there are those who read with interest these posts and gratefully exploring the links. A couple of wishes from me for a long, healthy and successful life You've earned, prodoljayte - and that light will very much alive

Джим · Post by **Джим** » Sat Oct 03, 2009 0:24

Not so much decent that stops on a half-word is still more than decent.

Well, Yes. Well-here for example the second half of the word-innet is a lot of code, which is considered ready-to-use solutions. It doesn't seem so. In this case-a man who knows how to program in C++ can write anything, including a signal approximation of a smooth curve. In many cases, it is workable. (For example, if you need to somehow recover the clipped sound). Maybe it will hold up in the case of noise (aka random), but in General, the sound -- the algorithm is shown in musicdsp.org it is not acceptable. Though cut.

Ie, criticism is understandable. But I have an alternative vision -- when on the forum are about 2-3 people and begin to discuss the finer aspects of the problems, interesting only to them-when I see it in someone else's performance, it annoys me (it's not on this forum).

So go on, please.
And good luck. All.

Джим · Post by **Джим** » Thu Oct 08, 2009 23:10

Malenikii gift::
http://yehar.com/blog/wp-content/upload ... 8/deip.pdf
-- I think a very good article about interpolation, its application is sinc(x)=sin(x)/x and how to choose the right solution.
And in General, the site http://yehar.com/ not bad, seems very; although, of course, not everything written there is suitable for processing sound -- which, however, always obvious from here is the author of symbols -- for example, a Hilbert transformation begins to work (according to the attached pictures) anywhere from 500-600 Hz (at FD==44.1 KHz), which is not good-but still interesting.
Good luck.

Волутар · Post by **Волутар** » Fri Oct 09, 2009 6:33

Jim, thank you so much for the link. Be sure to study.

About links in the algorithm on musicdsp, saying that it is not acceptable for music. Question - what is acceptable? To make the music "better"? Of course not! This banal Flanger effect, stick it selectively on individual sounds and not for the whole composition.
And if we limit ourselves to a simple delay (to emulate miasnoy delay HRTF), and in this case, nothing particularly bad will happen. In any case, emulation of 3D sound any sound affect its quality. Always. For spectral distortion, and all sorts of EHI in any way have place. In this sense, the slight distortion associated with fractional delay, will be absolutely minimal part in the overall picture of transformation.
Plus filtering itself contributes its delay and phase distortion into the sound. Of course... There is the idea that the overall sound was acceptable and in its way effective.
To apply these filters to any pre-recorded signal (not noise) I have not tried. If such a need arises, not at the beginning.

I still have not yet invented an elegant solution to separate settings for left/right (only where necessary). To make it convenient for editing and for rendering.

Джим · Post by **Джим** » Sat Oct 10, 2009 22:10

Волутар wrote:About links in the algorithm on musicdsp, saying that it is not acceptable for music. Question - what is acceptable? To make the music "better"? Of course not! This banal Flanger effect, stick it selectively on individual sounds and not for the whole composition.

On musicdsp well (probably) everything. Most importantly, not to do so simply copy/paste from there-it is necessary at least to understand (sorry, I'm a terrible bore). By the way, if all that is written there, really would have worked -- everything is already buried would Adobe autdition, and Sound Forge. Would be left alone Audacsity (or whatever it is?).
Yet, the attention-there (and only there) it is unreasonable use of the float. Operation double take MORE TIME!! At least on x86/x64. But bathe in precision is not necessary (well, almost). (By the way, integer operations require much more time than floating -- i.e. it is better to write in C/C++ double than in assembler, using 32x32=64 and add with carry 64-bit on Core I7 -- double wins by almost 2 times, on AMD K8-1.5).

Волутар · Post by **Волутар** » Sun Oct 11, 2009 13:54

Yes, stick float was effective 5 years ago, and int - all 10-20. In modern times, stick float can be effective only with SSE optimization (which I will be unlikely to do). When a direct calculation is not double brake. There is no doubt. I have all the basic calculations and go double.

Джим · Post by **Джим** » Tue Oct 13, 2009 1:13

By the way, about music.
I never was able to deal with OpenAL (see in search), but I think with the exception of rendering to a file (which is not obvious -- is there no -- but it seems You don't care) technology gives you all what You are fighting (zaoodno, a good way to forget about waveOut). Ie there, among other thingsyou can create a lot of single-band sources in 3D, and even make them move. With the participation of DSP's zvukovojj or programmatically -- it itself will determine (I on the first, and, alas, today is the last impression).
the rights of ideas.

Волутар · Post by **Волутар** » Tue Oct 13, 2009 9:44

OpenAL is just multiplatforming alternative to replace DirectSound and waveOut. She didn't give me. Channels to mix I do know how to delay non-multiple sampling from it I would not do. What I'm "stuck" I absolutely will not help to solve it. If after all these obstacles will be overcome and then decide to make a program multi-platform, then OpenAL can and will fit.

Джим · Post by **Джим** » Tue Oct 13, 2009 22:53

Valutar
AL only Open when I came (on a pie style) considered the possibility of "manufacturing" "broadband" sessions due to the motion of sources in space. Good, if worse there -- I don't have enough time with it to fully understand. But, in my opinion, if someone wants to play -- like a instrument, it appears, is not worse (now and in the future) than Direct Sound.
Well, it's our job to pay attention (experience shows that not all things are well-known -- well-known

).

Тим · Post by **Тим** » Mon Dec 07, 2009 18:51

Wow, well done, Valutar!
Actually, I came here by chance, read up on modulation noise for its optical problems, but your app is impressive! Thank you!

Волутар · Post by **Волутар** » Tue Dec 08, 2009 10:23

Yes, in fact, nothing special I did not. It is strange that previously did not... All music probably prefer.

I stuck with the easy view and edit stereo-separated parameters... unfortunately...

If the filter is divided into two parameters (for the left and right), then you need them to visually distinguish... but how? The bar crooked. Flowers - then it may be quite illogical: filter1 left - blue; right filter1 - red.
Although, it's probably the only viable option - the user will choose all these colors.

And stereo separation, as a rule (presumably) need for a small number of filters, so much color variety it, on-idea, should not create.

In this case, is to come up with new interface the top right of the box taking into account the specifics left/right. And there is not far to emulating HRTF.