Hello every one....

I have a problem of synchronizing my patch..is this is the right term to use.

When I run my patch the first time I have a result X, and when I run it the
second time I have a result Y, and the third time i have again the X result
and then the Y result and so on....and i cannot really figure out what i have
done wrong...

The main idea of my patch is:

[inlet~]
/ \
/ \
[compute spectra~] [compute pitch~]
| |
\ /
\ /
[compute jitter~]
|
[outlet~]

One idea that I had is that maybe my patches [compute spectra~] and
[compute pitch~] take different time to produce an output so I will have
to use a delay...but how can I count the time each one needs??

But is this the problem, or it has to do with my cpu? or the buffer that i use??

Hello every one!!!!!

So, a small intro... what i try to do it to compute the harmonic peaks of my
input (an "aaaaa" vowel)....

What i do is : my input-> hanning -> fft() -> abs() -> 20*log10()
and from that result i want to see where i have peaks....

I found the object peak~ witch takes a signal input and gives me a signal output...
really what i need...but the big big problem is that it only gives me 5-12 peaks in each processed block, and also not all the peaks that exist between its results,
(I hope i am clear). I looked for the help of the object, or the source code, but i did
not find anything.

What i want to do is to find around the first 160 peaks of my input signal and had
a correct output as in a format like the output of [peak~]....

Is there any ideas????

Hello to every body....

So, i have a txt file as input, and i would like to know, if there is an
object like [soundfiler~] to give me the lenght of the txt input file.
Or how i can compute it...something like :

until(EOF)
{
i++;
}

thank you

I have an array with the magnitude of the input signal.

How can I sum the values??? I cannot use the [sum] object because I have signal values...any ideas???

Good day to you all.

I'm trying to implement the NNE algorithm with pd.
NNE stands for Normalized Noise Energy, and is a ratio of the energy of the noise
present (above 1 kHz) in the vocalization to the total energy of the
signal (dB). This is all the information I have....I think it is similar to SNR but I'm not sure....

This is what I have already implemented....just until the computation of the energy
of a signal, and still not sure it is write. I do not know how to go on, and I'm a bit lost.

Do you have any ideas of what to do??

http://www.pdpatchrepo.info/hurleur/wav_1024samples_50%_overlap_Hanning.pd

Hi all,

This should be the easiest thing in the world, and it might seem to you very elementary,
but I'm really new with this pure data stuff (and a bit bloond - not much of advantage)
and I cannot figure this out.

I'm trying to read a wav file and put it into an array (fine with that), and then make fourier
transformation using a 40ms window with 20 ms shifting, and then compute the energies
of each window and store them in another array....

Can anybody please help me with this....give me some advice, hints, anything???

I would be grateful....

http://www.pdpatchrepo.info/hurleur/test.pd

Hello to everyone,

I'm very new in using pd and linux, and at this time I have a problem in installing
zexy 2.1. I downloaded zexy, read the "readme.txt" file and followed the steps.

autoconf seams to work, but with "./configure" I get this message

..........
checking "pd\>=0.37"... yes
checking "zexy-version"... 2.1
checking "parallel-port"... yes
configure: creating ./config.status
config.status: creating Makefile
./makesource.sh: 15: Syntax error: "(" unexpected

has anyone any idea what's going wrong???

(sorry for my english )

Just to get this right....if I use [block~ 2048 1 1] this means that my computation starts in a random position of this block?? I thought that block object just
increased the incoming block size, just splitting my input in bigger frames...I had
no idea that is a "random" object.

I tried your suggestion, which if I may say sounded great, but unfortunately it
does not work. In my patch I also use 50% ovelap.
So what I did after I read your idea was to replace [block~ 2048 2 1] with

[X]
|
[metro 50]
|
[switch~ 2048 2 1]

And it immediately crashed PD. So then I just eliminated the overlap, it run for some seconds and it crashed again...

Honestly I have no clue how it might be fixed or even what is going on

They are in the same subpatch with the same blocksize...

But what i cannot really understand is why I get two different results for the same input....do you have any idea what that could be???

Yes you are right...i use fft with 2048 block size.
All the objects that I use are ~ objects, how can I give a bang as an outlet?

How do I know that the output is not garbage?? and that the subpatch is actually finished?

Nice to have these.....thank you very much

Guys thank you very much for your help!!!

(i could not make it work with the soundfiler object ~ my blond hair problem,
but the patch you gave me is just fine!!!)

Thank you!!!!

I upload the patch that actually worked.....
Thank for your help and your time.....

http://www.pdpatchrepo.info/hurleur/compute_sum.pd

I will try it and i will tell you...
thank you for your help!!! (I'm very new with pd so....)

It is not that I'm worried, I do not get results because my pc "dies" when I use the fexpr~ object....
i also used [expr~ sum($v1)] but I get an error "expr: sum: need a table name"....
So I really do not know what to do. And it is very important for me to have the sum of the buffer each time it is filled.

The problem is that I have signal values, and each time i do that (using a for loop) i get different results. Probably because i cannot synchronize my input before it changes. and if i use [fexpr~] it is very cpu consuming....

Thank you very much!!!

I'll try it right away!!!

Thanks a lot again!!!