Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
Is it that you want to recognize a human voice?....In any case neural network has to be trained with sounds from various sources (training data) and the classify the sounds from different sources based on the training data accordingly. This is how you have to perceive your project. Specific doubts will be easier to solve.
Basically i have to recognise different kind of sounds i.e. door bell,mobile phone,ringing alarm etc using neural networks.
i want to implement this in matlab .... can u help me in implementing this??
MATLAB is easy to use for Neural Networks.
But first you have to define the number of neurons and no of layers you want to use.
these are simple equation which update its weights.
try to read about neural networks then explain your problem please.
my project topic "Speech recognition using neural network".
can any1 provide some documentation so that i can understand the topic very well.
I hav some doubt about the topic
1. what will be the no. of Input Nodes in the neural network? and what data it will contain
2. what will be the no. of Hidden nodes? and what data will it contain
3. Can a single layer replace the multiple layers on the neural network
3. Basic working of the network
4. what will be the voice features and how will be the training done in neural network
5. no. of bits it will operate on?
6. structure of the neural network used in the system
7. How many set of input values required?
There are two methods.
1) Speaker dependent recognition
2)Speaker independent recognition.
I don't know anything about speaker independent but in speaker dependent you have to find take atleast 10 samples of a single word (say "one") and then you have to find the melcepstrum for each and you have to give this as input to the neural network. You can use the neural network toolbox. There is also another separate window for pattern recognition in matlab. You can use that too.
Note: If you don't get accurate results then you have to increase the number of samples for every single word (say "one", "two" etc). The accuracy will keep on improving as you keep on increasing the number of samples.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
