| Sat, 03 Dec 2011 17:19:43 +0000 |
Andrew N Robertson |
hackday code is added.
|
| Wed, 30 Nov 2011 11:48:35 +0000 |
Andrew N Robertson |
follower has confidence measure now just using best match. Difficulty in visualising the speed likelihood fn
|
| Sun, 27 Nov 2011 21:56:19 +0000 |
Andrew N Robertson |
working well. Checking in before creating diagrams
|
| Fri, 25 Nov 2011 16:52:33 +0000 |
Andrew N Robertson |
real-time and offline version working well
|
| Tue, 22 Nov 2011 13:30:02 +0000 |
Andrew N Robertson |
offline method is now working, linked via pointer in BayesianStructure. Progresses using the time played so could be used for sequential offfline analysis
|
| Fri, 18 Nov 2011 11:56:32 +0000 |
Andrew N Robertson |
added in a mode to use either integrated estimate or MAP estimate in tempo process. Decreased nooise param from 0.02 to 0.03 so observations given more weight
|
| Tue, 15 Nov 2011 20:06:55 +0000 |
Andrew N Robertson |
Fixed the ordering of the note matrix when loaded so it has a double check feature that they are ordered min to max. Prevents draw problems down the line
|
| Tue, 15 Nov 2011 19:35:05 +0000 |
Andrew N Robertson |
reorders the MIDI matrix once loaded from all channels, chop the beginning now less relevant - check how this might work...
|
| Tue, 15 Nov 2011 13:10:12 +0000 |
Andrew N Robertson |
looking at offline version - need to do updateBestEstimate routine which presently only works for realtime
|
| Thu, 10 Nov 2011 17:29:45 +0000 |
Andrew N Robertson |
Checking to allow offline version
|
| Wed, 26 Oct 2011 14:25:46 +0100 |
Andrew N Robertson |
gave prior to tempo so it prefers original speed
|
| Tue, 25 Oct 2011 15:01:52 +0100 |
Andrew N Robertson |
fixed bug where the full midi file was sometimes not drawn (since minNoteIndexToPrint) was at the max and did not decrease properly in draw routine
|
| Mon, 17 Oct 2011 22:22:55 +0100 |
Andrew N Robertson |
added in the best estimate of error between the observed midi note and the aligned midi track.
|
| Tue, 23 Aug 2011 11:20:44 +0100 |
Andrew N Robertson |
looked at tempo process, likelihood function changed and improved. Now drawing using constrained vector function. Good working version.
|
| Fri, 19 Aug 2011 19:42:01 +0100 |
Andrew N Robertson |
added confidence measure to each played event and associated match, but not really working yet
|
| Fri, 19 Aug 2011 16:38:30 +0100 |
Andrew N Robertson |
added purple where notes have been seen - lets you see what updates have been used. Also the chopping of midi files to the beginning was introduced recently, so when they load, you chop any white space at the beginning, then use first note to launch.
|
| Fri, 19 Aug 2011 15:53:04 +0100 |
Andrew N Robertson |
optimised draw function in dynamic vector class. Added Gaussian lookup but not yet used.
|
| Fri, 19 Aug 2011 02:36:34 +0100 |
Andrew N Robertson |
added speed prior setting. needs some cleaning
|
| Fri, 19 Aug 2011 01:26:40 +0100 |
Andrew N Robertson |
Added separate CannamMidiFileLoader class to handle the loading in.
|
| Thu, 18 Aug 2011 23:27:42 +0100 |
Andrew N Robertson |
Changed bestEstimate timing to work via time sent from Max not the elapsed time. This had caused some problems, but this version now working surprisingly well on MIDI files with variable timing.
|
| Tue, 16 Aug 2011 11:29:59 +0100 |
Andrew N Robertson |
start project
|
