Home My Page Projects MPTK: The Matching Pursuit ToolKit
Summary Activity Forums Tracker Lists Docs News SCM Files
Close InriaForge Scheduled outage on Tuesday October 30, 2018, from 2PM CEST

Forum: help

Monitor Forum | Start New Thread Start New Thread
RE: basic questions: anywavehilbert, nonnegative weights [ Reply ]
By: Graham Coleman on 2014-06-28 15:22
[forum:148677]
Answering my own questions:

1. Using Gabor atoms as an example, using analytic atoms allows summing the real and imaginary parts to get atoms with arbitrary phase. Doing this with arbitrary real atoms is analogous, allowing arbitrary phase shifts to your atoms.

2. This does not seem to be possible. It could be because of the highly optimized MP routines, or it could be because no one needs non-negative decompositions on real audio.

Graham

basic questions: anywavehilbert, nonnegative weights [ Reply ]
By: Graham Coleman on 2014-06-20 13:10
[forum:148650]
Greetings,

I wanted to do some basic experiments with MPTK, and everything seemed to have worked so far with the latest Windows binary (0.6.1). I created a custom anywave table from sampled waveforms and the results seem reasonable.

1. What are the "anywavehilbert" atoms? In the example dictionaries in "mptk\reference\dictionary", an anywavehilbert block is invariably paired with an anywave block, but I haven't seen them explained in the manual anywhere. Are they necessary?

2. Does MPTK support matching pursuit with non-negative weights?

best regards,

Graham Coleman