Mercurial > hg > constant-q-cpp
diff yeti/cqt.yeti @ 11:d7af615c2cc1
Comments
author | Chris Cannam <c.cannam@qmul.ac.uk> |
---|---|
date | Fri, 25 Oct 2013 17:39:28 +0100 |
parents | 3dd3008f3b19 |
children | e2372a642b45 |
line wrap: on
line diff
--- a/yeti/cqt.yeti Fri Oct 25 16:57:04 2013 +0100 +++ b/yeti/cqt.yeti Fri Oct 25 17:39:28 2013 +0100 @@ -32,6 +32,8 @@ kdata = cqtkernel.makeKernel { sampleRate, maxFreq, binsPerOctave }; + println "atomsPerFrame = \(kdata.atomsPerFrame)"; + streams = manipulate.duplicated octaves str; //!!! can't be right! @@ -52,6 +54,14 @@ done frames; done [0..octaves-1]; + // cqblocks is now a list<list<matrix>>. Each top-level list + // corresponds to an octave, from highest to lowest, each having + // twice as many elements in its list as the next octave. The + // sub-lists are sampled in time with an effective spacing of + // fftSize * 2^(octave-1) audio frames, and the matrices are row + // vectors with atomsPerFrame * binsPerOctave complex elements. + + //!!! The comment below isn't true -- this is based on traditional // cqt -- the atom hop structure for this version is different // though so review later on @@ -72,7 +82,6 @@ // For each frame in the lowest octave println "prepared block list"; - cqblocks; );