STFT Spectrogram - 50% Overlap

[tinnitus]

hello Dan

Could you tell me in wich training you are ?

@+

Tinnitus

[dan07]

Hello...

Sorry, I did not understand your question, "which training?"

Thanks

Dan07

Message Edited by dan07 on 01-30-2009 08:53 AM

[tinnitus]

sorry,

what is tou study level, technician, engineer, at school or at work

best regards

[dan07]

Hello,

I am a PhD student on medical sciences, but I have a great interest in data analysis. I am improving my Labview skill, but sometimes some basic math concepts are missed and I don't know how to solve the problem.

Concerning my answer about how to match the 34 values for mean and stdvs and the 34 spectrogram rows (after the removal of first and last row), may I accept as true that the first mean and stdv value will correspond to the first row of the spectrogram? In other words, the data used to generate the first row of spectrogram is the same that used for first mean and stdv results?

Thanks for your great attention

Dan07

[dan07]

I made a mistake on the last message:

"Concerning my question" its the correct.... I wrote "Concerning my answer"

Dan07

[tinnitus]

---

I am a PhD student on medical sciences, but I have a great interest in data analysis. I am improving my Labview skill, but sometimes some basic math concepts are missed and I don't know how to solve the problem.

---

isn't it a little bit dangerous for patients ?

here the explanation see on attached jpeg,

1 is window used by stft

2 window used with for loop 

That is why you have 2 more rows in final array , the first and the last ...

See you 

Tinnitus

[dan07]

Thanks for explanation. So, I can remove the first and last rows of the spectrogram, and match the 34 rows of the spectrogram with the 34 means and stdvs values: first row of spectrogram matchs with first mean value, second row of spectrogram matchs with second mean value..... all right?

Thanks

Dan07

[tinnitus]

yes...

[dan07]

Hello...

After all your help, the code its working properly now... the only doubt that I still have concerns the huge difference between the absolute values showed by specgram (matlab) and spectrogram (labview). The normalized values (%) are exactly the same, but the absolute values returned by specgram (matlab) are much greater (40x greater, or sometimes 200x greater). Do you have a idea about what might be happening?

Thanks

Dan07

[tinnitus · Accepted solution]

hello Dan,

did you compare graphic or array values .

did you compare right after specgram function ?  because if i remember well there was a CxC* in matlab code

did you implement same in labview code ? conjugate block is in numeric->complex->...

tell if this helps

@+

Tinnitus