wav file > FFT > peak point

[zeldAIS]

Hellow. I'm a relatively complete newb at LabView and in need of immediate help in making a vi. I just stumbled upon here when i decided to research at the internet on LabVIEW for the programming part of my thesis (I used to endlessly search at the help provided with the program of the LabView, tried to use some hits, stare and still can't get too far)

My thesis is about determination of maturity levels of young coconuts through acoustic response. Basically it will get tapping sound signals from young coconuts, then it will get the FFT and specify what maturity it is (malauhog(mucous-like); malakanin(cooked rice-like); malakad(leather-like)). below is the flowchart for the algorithm I made.

I think i got the latter part where there is a selection. Kinda modified it at the unfinished vi i've done so far. Uploaded it in here. I still don't get how i will be able to make the user input a recorded wav file > FFT it > get peak point.

My adviser (who is also relatively new at LabView) have switched the programming part of my thesis from Digital Signal Processor (as i'm also going nowhere there even if i have the flowchart and got the FFT signals of the tapping sounds; and also - it only works at windows 98 os) to LabView just this month (which explains my uber newbiness to it)! i need immediate help since i have to finish the program before tuesday (2007-05-22) T_T ...and sorry if i posted kinda late (as i never did knew there were forums like this - should've googled long time ago) and sorry if i posted at the wrong subforums.

In desperate need of immediate help. Thank you in advance.

Edit - i tried also using the parts explained from this thread.

[Anders Björk]

First of all remove the Fourier transform vi there is no need for that. The tone measurement express vi make the complete work for you. You only need to wire input data (wavfile data/AI) and specify a approximate frequency for the peak your looking for (possibly specifiy some other parameters).

To read a wave file you find them in the palette Programming\Graphics and Sound\Sound.

Rather than attaching VI paste a screenshot of your diagram, you will get a better respones, fast for us to see your problem.

[Gary Rubin]

QUOTE(zeldAIS @ May 18 2007, 08:54 PM)

My thesis is about determination of maturity levels of young coconuts through acoustic response.

I think this wins the prize for the most interesting/unusual application of Labview! :thumbup:

Hmmm... It got me thinking - if I had a PDA with Labview on it running an application like this, maybe I could finally figure out how to pick a good melon at the supermarket...

[zeldAIS]

QUOTE(Gary Rubin @ May 19 2007, 08:05 PM)

I think this wins the prize for the most interesting/unusual application of Labview! :thumbup:

Hmmm... It got me thinking - if I had a PDA with Labview on it running an application like this, maybe I could finally figure out how to pick a good melon at the supermarket...

hahaha Yea, this is quite an interesting topic which my father wanted to work on way back and just passed it to me (my father is a university professor btw). The data I acquired seems reliable in the mucous-like and cooked rice-like - but there seems a lot of noise at leather-like. My adviser and my father said that I just use recommendations for further study since this is just an undergrad thesis

QUOTE(AndersBjork @ May 19 2007, 05:03 PM)

First of all remove the Fourier transform vi there is no need for that. The tone measurement express vi make the complete work for you. You only need to wire input data (wavfile data/AI) and specify a approximate frequency for the peak your looking for (possibly specifiy some other parameters).

To read a wave file you find them in the palette Programming\Graphics and Sound\Sound.

Rather than attaching VI paste a screenshot of your diagram, you will get a better respones, fast for us to see your problem.

Thank you very much for replying :worship: I'll prolly mostly redo the block diagram with your suggestion.

hmmm...is there a possibility that the tone measurement will output the FFT of it? I've been trying hours trying to figure that out - but no luck. cause i noticed that there is an Input Signal and a Result Preview when you select the properties of the tone measurement. I will most likely be using the amplitude since i need the peakpoints. If it is possible - kindly please post how asap. as i'm desperate

thanks again in advance

[Anders Björk]

QUOTE(zeldAIS @ May 20 2007, 12:47 AM)

hahaha Yea, this is quite an interesting topic which my father wanted to work on way back and just passed it to me (my father is a university professor btw). The data I acquired seems reliable in the mucous-like and cooked rice-like - but there seems a lot of noise at leather-like. My adviser and my father said that I just use recommendations for further study since this is just an undergrad thesis

Thank you very much for replying :worship: I'll prolly mostly redo the block diagram with your suggestion.

hmmm...is there a possibility that the tone measurement will output the FFT of it? I've been trying hours trying to figure that out - but no luck. cause i noticed that there is an Input Signal and a Result Preview when you select the properties of the tone measurement. I will most likely be using the amplitude since i need the peakpoints. If it is possible - kindly please post how asap. as i'm desperate

thanks again in advance

Well put a Express for Spectral measurements in parallell with the tone-measurements or edit the tone measurement vi... I guess the first alternative is most easy. That extra computing will probally not be a problem in your case.

[Anders Björk]

QUOTE(AndersBjork @ May 20 2007, 03:16 PM)

Well put a Express for Spectral measurements in parallell with the tone-measurements or edit the tone measurement vi... I guess the first alternative is most easy. That extra computing will probally not be a problem in your case.

Adding a comment on your evaluation, I would use principal component analysis or partial least squares to analyse the spectra or selected regions around the peaks. I have done similar thing since 1997. You should get a better estimate of the "condition" if you use a "weighted" average.

[aka]

hej,

a fast way to do your DSP would be to call Matlab from Labview (that's if you're familiar with Matlab). then you can load the wav file , do the FFT and analysis in Matlab (this kinda makes Labview obsolete though...!)

otherwise,

in Labview to :

load a wav: Snd Play Wave File.vi

FFT: under Functions there are vi's to do FFT

bandpass filters: see under Functions\analyze\signal processing\filters

LEDs: control the LEDs with outPort.vi

good luck

ps. if you have the time please tell me why it is that getting just the peak at a certain frequency range will tell you something...why not energy in the octave/third octave bands. how do you excite the coconut?

alex

[zeldAIS]

It's been awhile... and i thank you guys deeply for the help.

anyway, my panels were not satisfied with what i made and they asked to revise the vi i made. They asked to change the input instead of an edited WAV file (that had been noise filtered and normalized) to a sound input of a microphone that will determine the level of maturity of the coconut.

aka (alex) said - i used a filter in the input of the sound. (and alex, i "excite" the coconut by tapping the fruit)

here are screenshots of the vi i revised

the problem now is that it always gives me an "out of range" when i know for sure that the coconut is "malauhog" or mucous-like and the other levels of maturity gives me an "out of range" value also....*sigh*

any ideas how i should solve this? please help....