samplomatic

The samplomatic gathers sounds from an incoming audio stream based on an amplitude threshold. Once a sound is captured, its frequency spectrum is analysed by means of Fast Fourier Transform. This frequency spectrum is made available in the filename of the wav file that the samplomatic writes.

So the samplomatic listens to your soundcard and makes little wav files from every audible sound it hears. You can tell a bit about the characteristics of the sound by the filename of a given wav file. You can tune the threshold to get more or fewer samples, and you can tune the "fuzz" to get longer, more intelligible samples or choppier samples.

The original purpose behind making the samplomatic was for it to automatically gather sounds for the syncotron to use. However, it has performed nicely as an automated way to gather sounds for use in other sequencing software, such as the popular ACID.

versions

Version 1.0 is the one that made its debut at BM2K. It depended on the Portable Object Compiler. It was not easy to build.

Newer versions are beta because they are not as well tested, but they should be fine, since the improvements are mostly just things that make the samplomatic easier to build.

Please let me know how the beta version builds and works for you. I don't recommend the stable version, since it was made to be built on my machine.

requirements

• Objective-C compiler (like gcc)

  Since the samplomatic is written in Objective-C, you will need a compiler for Objective-C. Fortunately, gcc includes an Objective-C compiler.

  If you're using a packaging system like RedHat's RPM, make sure that the compiler package that has Objective-C support is installed. e.g., egcs-objc or gcc-objc.

• FFTW library

  The samplomatic uses the Fastest Fourier Transform in the West to do the spectral analysis. You should check it out if you haven't heard of it. FFTW taylor makes an algorithm just for your own computer after testing its strengths and weaknesses.

  (I'm using the RedHat RPMs, which install both the single and double-precision versions of the library by prefixing the double version with the letter 'd'.)

• sndlib, the sound programming library at CCRMA. I make a symlink from "sndlib.a" to "libsnd.a" so that I can say "gcc -lsnd".

  In the future the samplomatic probably will not depend on sndlib, since all it uses it for is audio capture and wavfile writing.