I was wondering if anyone could provide guidance on audio analysis. Analyzing waveforms or pitches and speech is what I'm looking at. I am pretty much open to using any C-based language (C++,C#,Java), and if that's not possible than maybe a new language. Does anyone have any tips on where I should start?
+2
A:
That's pretty broad, but the first thing to do with a problem like this is to figure out some useful search terms. You could start with the wikipedia category listing for Signal Processing for example.
It's likely you'll find Fourier Analysis worth investigating - that might provide a good starting point for learning. Armed with that you can learn how to extract particular frequency components from a complex waveform.
You might also like to check out other questions on Stack Overflow
Paul Dixon
2009-02-20 22:37:26
A:
You could look here for a FFT library that interfaces to C (and C++). AFIK there's language bindings for heaps of others too.
I suggest that since you ask this question you know S.F.A about DSP and should begin with some elementary reading. There's a lot of knowledge you'll need to glean before you're able to code an audio processor (even using available tools). Paul has provided some links to SO, but get on the search engine and look.
I sound like a complete prick, but without the mathematics backing you'll dig yourself into a hole and possibly start getting results that look OK but are wildly inaccurate.
I'd recommend Matlab as a starting point because it has all the FFT code as well slow DFT code in it. You'll be easily able to graph results and change params on the fly. You should even be able to learn the mathematical basis by having a good reference and feeding code into Matlab. There's a 15 day trial and last I checked (while studying my Ph.D ages ago) the student version was very cheap.
Adam Hawes
2009-02-21 00:57:31