Detecting peaks in realtime from a microphone

Hi to all.

I start by apologizing for my newbyosity in this field of c++, that is audio tracks. I understand, however, the principles behind sampling and sound recording issues in a computer.

I put in open terms. I am developing a software that needs to be able to detect when the microphone input sound reaches a certain level.

I see the API native functions for the waveforms, like waveIn..., waveOut... and so on, but I don't know what to choice to get sound in input in realtime.
The problem is not so difficult, if I could access the values of incoming samples, even one each X samples, and compare it with a percentage of the maximum value allowing by input device dynamics. The fact is, I don't know how to do, and if it is possible with the waveform functions I named.

Sorry for the problem-specific question, I appreciate any help.

Thanks in advance. Oh, and happy 2010!!!

Reading the peaks is the trivial part. Just iterate through your audio buffer and keep the highest value. The tough part is implementing a routine to sample the audio. Check out some of the classes here:

http://www.codeproject.com/KB/audio-...eo%20-%20Audio

and here:

http://www.codeguru.com/cpp/g-m/multimedia/

I've looked at many and used ideas from several to come up with an implementation right for my own purposes.

I viewed some of these, and found some interesting. Still reading, there are many articles. Thank you for the addressing.

But, you said the difficult part is to sample the audio; there are API functions that record the input audio to save it in a file. Could that data be read or interpreted as sampled audio in RT, block by block, instead of sending them to a file on memory? Do I really need to do the sampling on my own?

I mean, when I have a HWAVEIN received with WaveIn function, could I navigate the data related to it as an "array of samples", or something like that?

Windows API adds some simple wrappers for all the necessary handling of simple playback (sndPlaySound() ) but it obscures all the internal stuff from the user. I don't know of any way to get at the actual buffers. Nonetheless, recording is a different beast altogether. I don't know of any other way than to: acutally define a callback, threadproc, or message handler called each time the buffer is filled by the sound hardware; create the buffers; then continuously manage the buffers (save to file, monitor, etc.) each time the callback signals a full buffer. It's an ugly process and I understand that it is a little better using DirectX sound APIs, but I've never gone that route.

You've been very clear. Thank you again, I will try something out of your suggestion.