Forums

Peak Isolation Algorithm

Started by Sattu April 19, 2006
One of the good features to recognize the voice gender is the
periodicity of residual autocorrelation function. To implement the
algorithm, We calculate the distance between the first and second  peak
of autocorrelation for each frame , average it all over the nonzero
frames and take it as a feature of gender voice. Needless to say, using
autocorrelation of residual can be identified as linear method using
the vocal generator behavior. To train the classifier, we average the
peak isolation of males and females separately, and denoted as p_man
and p_woman. To classify the input voice, these are the steps:

1. Find peak value for input voice
2. check whether is closer to p_man or p_woman, output male or female
accordingly 

Can I have MatLab Code for this??
Thanks in Advance

"Sattu" <satyendra.choudhery@gmail.com> wrote in message 
news:1145419464.326522.302970@i39g2000cwa.googlegroups.com...
> One of the good features to recognize the voice gender is the > periodicity of residual autocorrelation function. To implement the > algorithm, We calculate the distance between the first and second peak > of autocorrelation for each frame , average it all over the nonzero > frames and take it as a feature of gender voice. Needless to say, using > autocorrelation of residual can be identified as linear method using > the vocal generator behavior. To train the classifier, we average the > peak isolation of males and females separately, and denoted as p_man > and p_woman. To classify the input voice, these are the steps: > > 1. Find peak value for input voice > 2. check whether is closer to p_man or p_woman, output male or female > accordingly > > Can I have MatLab Code for this?? > Thanks in Advance >
Ask one of your classmates - they may be willing to give you their code for a fee.
if you are ready to give me code for this then I cangive you your fee
;)
Ne way can you help me regarding this.??
Its kinda urgent.

"Sattu" <satyendra.choudhery@gmail.com> wrote in message 
news:1145420244.929655.44260@i39g2000cwa.googlegroups.com...
> if you are ready to give me code for this then I cangive you your fee > ;) > Ne way can you help me regarding this.?? > Its kinda urgent. >
Hello Sattu, You have a sound set of steps written out to see which population a voice belongs to. Is the difficulty actually writing the code? Isn't one of the goals of the class being able to take a set of steps and actually PROGRAMING the computer to work out an answer? If I were to do your work for you - how do you intellectually benefit? Part of the learning process is had by doing. Since I spend some time teaching, I can't help you cheat. This forum serves many purposes, but it should not be used for cheating. If you have a specific problem, then many here can help, but try to do your own work. Clay
if you are ready to give me code for this then I cangive you your fee
;)
Ne way can you help me regarding this.??
Its kinda urgent.

Sattu skrev:
> One of the good features to recognize the voice gender is the > periodicity of residual autocorrelation function. To implement the > algorithm, We calculate the distance between the first and second peak > of autocorrelation for each frame , average it all over the nonzero > frames and take it as a feature of gender voice. Needless to say, using > autocorrelation of residual can be identified as linear method using > the vocal generator behavior. To train the classifier, we average the > peak isolation of males and females separately, and denoted as p_man > and p_woman. To classify the input voice, these are the steps: > > 1. Find peak value for input voice > 2. check whether is closer to p_man or p_woman, output male or female > accordingly > > Can I have MatLab Code for this?? > Thanks in Advance
People here don't do your homework for you. We try as best we can help you do your own homework. There are numerous reasons for such "grumpy" attitudes. First, lots of people here do teaching for a living. Because of that, we don't help students "freeload". It's a matter of principle. Next, most of us have worked very hard for very long to learn whatever we know. If there is a consensus on any matter whatsoever amongst the regulars here (which I doubt there is), it would be that what DSP is concerned, hard work pays off. Last, there is the chance one of us might get you as a colleague one day. If that happens, we would appreciate if you know the basics your job. If you don't, please join one of the competitors... Rune
"Clay S. Turner" schrieb 
> > > > [Re: Peak Isolation Algorithm] > > > > Can I have MatLab Code for this?? > > Thanks in Advance > > > > > Ask one of your classmates - they may be willing > to give you their code for a fee. >
Or he may buy the latest version of dBase [1], which is mindboggingly easy to program: # set errors off # do what I am thinking SCNR Martin [1] From the introduction of the programmer's manual for dBase III