Processing DirectSoundCapture data
Author
Hi all,
I'm trying to use captured sound in my game, and I've managed to get sound data from DirectSound fine, but I don't know how to use the captured wave data.
Does anyone know of any good libraries/references for processing the wave data return in DirectSoundCapture buffers? I've looked online a lot but can't find anything useful. Ideally I'd like to be able to break down the sounds into frequency ranges and measure how much sound there is within those ranges - like a graphical equaliser.
Thanks for any help, Duncan
Author
Thanks for that :) I've found a library called FFTReal... though I don't really get how to use it, do I just bung the sound signal into an array, throw it in, and then something magically pops out? Has anyone used this lib before? The test samples are so wrapped in template garbage I can't see how to use it.
Author
Really I don't actually know how to deal with the bytes in a capture buffer, and I haven't been able to find any tutorials that don't do something as simple as just throwing it into a WAV file...
... does anyone know any tutorials that do something non-trivial with capture buffer data?
... does anyone know any tutorials that do something non-trivial with capture buffer data?
The core of any FFT library is the transform and inverse transform functions.
The transform, given a time-domain sample (a byte array of the audio data in this case), returns a frequency-domain sample (an array of imaginary numbers describing the amplitudes of the frequency bands and their phases).
The amplitudes of the frequency bands are essentially what you see in an audio frequency visualizer. The values and their sampling points need to be scaled up logarithmically, so as to weight the amplitudes of the different frequencies to a common scale for display (you know what I mean once you see the values unfiltered).
The inverse FFT reverses this process - it takes the array of the frequency band amplitude and phase information and combines the data to form a sample of sound.
If you want to do - for example - band filtering or boosting, first transform the sample to frequency data, then scale the frequency bands' amplitudes by your own multiplier array, and then transform the data back to a sample.
The transform, given a time-domain sample (a byte array of the audio data in this case), returns a frequency-domain sample (an array of imaginary numbers describing the amplitudes of the frequency bands and their phases).
The amplitudes of the frequency bands are essentially what you see in an audio frequency visualizer. The values and their sampling points need to be scaled up logarithmically, so as to weight the amplitudes of the different frequencies to a common scale for display (you know what I mean once you see the values unfiltered).
The inverse FFT reverses this process - it takes the array of the frequency band amplitude and phase information and combines the data to form a sample of sound.
If you want to do - for example - band filtering or boosting, first transform the sample to frequency data, then scale the frequency bands' amplitudes by your own multiplier array, and then transform the data back to a sample.
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement
