B4A Library FFT Fast Fourier transform library

Thanks Klaus ... I'll try to make a post in the forum (in the B4i wishlist forum)

 

@klaus : can we use this lib on a returned soundbuffer to display a waveform or frequency bars? (from the MIC or stream)

or will it be too slow?

 

Waveform no.

Frequency bars, I don't know, I have never tested it.
It depends also on the number of time signal items you want to take into account.

 

Using another library on this site somewhere to get the audio data in real time, I use this to get the frequency bars

Code:

Type Histogram(Values As List, Highest As Int, Inc As Int)
    Dim TimeMin As Double=1000000,TimeMax As Double =-1000000, FFTMax As Double ,  FFTLimit As Double, FFTThreshold = 0.8 As Double, N_Samples As Int , CursorScaleFreq As Double , FFTprev As Int 
    Dim FFT1 As FFT

Sub RecData2Histogram(SampleRate As Int, recData() As Short) As Histogram ', recData() As Double =  fft1.CopyArray(srcData)
    Dim tempHist As Histogram ,  soundD() As Double = CopyRecDataD(SampleRate, recData), N_2 As Int = soundD.Length * 0.5 ,FFTPeaks As List 
    Dim FFTReal(N_2) As Double, FFTImg(N_2) As Double, FFTAmp(N_2) As Double
    
    tempHist.Initialize 
    tempHist.Values.Initialize 
    
    FFT1.Transform2(soundD, FFTReal, FFTImg )
    FFTAmp = FFT1.ToAmplitude(FFTReal,FFTImg)
    PeakDetect(FFTAmp)
    
    tempHist.Inc = 1
    tempHist.Highest = FFTMax' TimeMax
    tempHist.values.AddAll(FFTAmp)
    
    Return tempHist
End Sub

 

Hi Klaus!! Let me know when you start to make the ios class of this project ... badly need it ...

 

Unfortunately you need to wait a bit, I will be on travel for one week.
I will have a look at it at the end of next week.

 

Thank you very much for the library and the test project posted.
Could you please help me figure out why I get this?

Code:

** Activity (main) Create, isFirst = true **


** Activity (main) Resume **


Start Recording


Sampling...


32768


AR state: 1


java.lang.RuntimeException: FFT: Real and Imaginary arrays are incorrect length!


    at anywheresoftware.b4a.agraham.fft.FFT.Transform2(FFT.java:253)
    at b4a.example.main._initcalulation(main.java:739)
    at b4a.example.main._btncalc_click(main.java:458)
    at java.lang.reflect.Method.invokeNative(Native Method)
    at java.lang.reflect.Method.invoke(Method.java:515)
    at anywheresoftware.b4a.BA.raiseEvent2(BA.java:175)
    at anywheresoftware.b4a.BA.raiseEvent2(BA.java:163)
    at anywheresoftware.b4a.BA.raiseEvent(BA.java:159)
    at anywheresoftware.b4a.objects.ViewWrapper$1.onClick(ViewWrapper.java:66)
    at android.view.View.performClick(View.java:4424)
    at android.view.View$PerformClick.run(View.java:18383)
    at android.os.Handler.handleCallback(Handler.java:733)
    at android.os.Handler.dispatchMessage(Handler.java:95)
    at android.os.Looper.loop(Looper.java:137)
    at android.app.ActivityThread.main(ActivityThread.java:4998)
    at java.lang.reflect.Method.invokeNative(Native Method)


    at java.lang.reflect.Method.invoke(Method.java:515)
    at com.android.internal.os.ZygoteInit$MethodAndArgsCaller.run(ZygoteInit.java:811)
    at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:627)
    at dalvik.system.NativeStart.main(Native Method)
java.lang.RuntimeException: FFT: Real and Imaginary arrays are incorrect length!

 

Hello and thank you for posting this code.
How is this implemented in a project?
I mean I would like to get the frequency of a tone recorded by the device mic.

 

Would the following do the trick?

Code:

Public N = 16384 As Int
    Public N_2 = N / 2 As Int
   
    Public soundD(N) As Double

    Public FFT1 As FFT
    Public FFTReal(N_2+1) As Double 'added +1
    Public FFTImg(N_2+1) As Double 'added +1
    Public FFTAmp(N_2) As Double

 

@padvou
Sorry for answering only today, but I was on travel for a week so no B4x.
The declarations must be:

Code:

Public N = 16384 As Int
Public N_2 = N / 2 As Int
Public N_21 = N_2 + 1 As Int
Public soundD(N) As Double
Public FFT1 As FFT
Public FFTReal(N_21) As Double
Public FFTImg(N_21) As Double
Public FFTAmp(N_21) As Double
' Public FFTPhase(N_21) As Double

for use with Transform2 and Inverse2.

 

@NewTechni
Be careful, the FFT library had been updated.
In the new version the array dimensions must be N / 2 + 1 !
Look above.

 

@klaus
Thank you very much for responding to me.
Could you please also tell me what this is:

What is the threshold for peak detection?

 

Hi, my name's Luis, i've been trying to use the FreqDetect.zip but i get the error:
"An error has occurred in sub: java.lang.RuntimeException: FFT: Real and Imaginary arrays are incorrect lenght! Continue?"

I also tried the FFT_Freq_Detect and i also get the same error...

This is the sound file that i've been using to check if the frequency is correct:



Does anyone have the same error? any suggestion? Thanks for any tip in advance

 

Hi again, i've solve my problem reading the past post from Klaus, just change it like in the picture.

Thanks Klaus and Graham, awesome work!!!

 

I have updated the projects in post #42 and #51.

 

Hi Klaus! Any news about the ios FFT Class? Can I help somehow?

y

 

I like your FreqDetect Zip, but I need to modify it somehow and I am not sure how.

I dont need graphs or anything, but what I do need is realtime monitoring of the mic for frequency peaks and amplitudes. Then I have a stored list with the certain frequencies and amplitudes stored, soon as I hit a match between stored list and FFT list, it fires an event.

Your FFT program can basically do that, but it requires "recording" which I dont want. I need realtime live monitoring so the instant the sound is heard, analyzed, then matched it fires an event. Event runs, and then it keeps on moving until its heard again.

 

Just tried the FFT detect and I dont think it works. Both of my phones keep showing the same 3 peaks, no matter if the room is silent or if I am playing something, plus I have to keep hitting record over and over. its NOT realtime!

any thoughts???

Peaks are 500.65, 899.1, and 1300.7 no matter what.

 

And how would it be for a Z transform?

 

You would need to write it.