Audio Feature Extraction in VHDL

[tarang]

audio vhdl

Hello folks,

I am a beginner to VHDL. I have basic knowledge of VHDL programming. I have to use VHDL for my final project now where i have to implement speech recognition using FPGA.

Starting first can anyone guide on how can i extract various features of a speech signal using VHDL. Any suggestion on book or any tutorials will be great full ...

thanx....

 

[ring0]

vhdl audio

I can't help you with your project, but I suppose that at the beginning you should pay more attention to speech recognition algorithms than to VHDL/FPGA. As long as algorithm is chosen and simulated, the implementation won't usually be a problem.

 

[tarang]

What i am asking is just a small push or something....

We have decided on the algorithm...which will use Hidden Markov Model (HMM) but this is the comparison part. I want to learn what steps should be followed in order to make the incoming voice signal through microphone comparable....

There are few sample pre recorded voice data that will already be available in the memory...we will extract features from that by using software such as MATLAB say, but the input voice signal has to go through hardware....and it need to be processed by the hardware...so i want to know how can i do that?

 

[FvM]

I basically agree with ring0's statement. The algorithm matters! Generally, you are able to implement any digital signal processing algorithm, that is avaiable with MatLab in FPGA. MatLab has toolboxes as fixed point DSP that directly support hardware implementation. There is some basic literature as Uwe Meyer-Baese DSP with FPGA dealing with implementation details as numeric presentation and arithmetics.

Considering the processing throughput requirements, audio DSP doesn't necessarily need FPGA, DSP processors and fast GP processors can also do the job.