| Fri, 27 Feb 2015 11:38:04 +0000 |
Chris Cannam |
Stats
memory
|
| Fri, 27 Feb 2015 10:44:13 +0000 |
Chris Cannam |
Another process wrapper
|
| Fri, 27 Feb 2015 10:05:04 +0000 |
Chris Cannam |
Fix off-by-one in durations
|
| Fri, 27 Feb 2015 09:29:37 +0000 |
Chris Cannam |
Merge from branch re-minimise
|
| Fri, 27 Feb 2015 09:25:23 +0000 |
Chris Cannam |
Expose distance scale parameter
re-minimise
|
| Fri, 27 Feb 2015 09:07:54 +0000 |
Chris Cannam |
Increase noise for compact types
re-minimise
|
| Fri, 27 Feb 2015 08:55:39 +0000 |
Chris Cannam |
Fix tests
re-minimise
|
| Thu, 26 Feb 2015 17:12:56 +0000 |
Chris Cannam |
Default min freq to 100
re-minimise
|
| Thu, 26 Feb 2015 15:57:50 +0000 |
Chris Cannam |
Use compact types everywhere by default
re-minimise
|
| Thu, 26 Feb 2015 15:57:32 +0000 |
Chris Cannam |
Default to something closer to the original MATCH
re-minimise
|
| Thu, 26 Feb 2015 15:55:52 +0000 |
Chris Cannam |
Fix an erroneous test
re-minimise
|
| Thu, 26 Feb 2015 15:52:16 +0000 |
Chris Cannam |
Normalised path costs should use a different type from un-normalised ones (because they are in a different range)
re-minimise
|
| Thu, 26 Feb 2015 15:51:50 +0000 |
Chris Cannam |
Print some info about scale range, so we can work out what scale factor to use
re-minimise
|
| Thu, 26 Feb 2015 12:26:33 +0000 |
Chris Cannam |
Fix typo
re-minimise
|
| Thu, 26 Feb 2015 12:19:17 +0000 |
Chris Cannam |
distance_t * diagonalWeight might not fit in distance_t; use pathcost_t for it. Also remove C-style casts.
re-minimise
|
| Thu, 26 Feb 2015 10:48:36 +0000 |
Chris Cannam |
Types.h -> MatchTypes.h
re-minimise
|
| Thu, 26 Feb 2015 10:38:59 +0000 |
Chris Cannam |
Fix tests
re-minimise
|
| Thu, 26 Feb 2015 10:28:23 +0000 |
Chris Cannam |
Code now builds, DistanceMetric tests fail
re-minimise
|
| Thu, 26 Feb 2015 09:55:28 +0000 |
Chris Cannam |
Some fixes (still does not yet build)
re-minimise
|
| Thu, 19 Feb 2015 17:17:20 +0000 |
Chris Cannam |
Toward allowing types to be small again. Doesn't currently build
re-minimise
|
| Thu, 19 Feb 2015 16:57:19 +0000 |
Chris Cannam |
Start using properly-named types
types
|
| Thu, 19 Feb 2015 16:45:42 +0000 |
Chris Cannam |
Introduce (though don't use properly) types header; get to build
types
|
| Thu, 19 Feb 2015 16:14:33 +0000 |
Chris Cannam |
Build with -Wconversion
tuning-rescale
|
| Mon, 16 Feb 2015 15:38:42 +0000 |
Chris Cannam |
This flag seems too problematic with the gcc builds people are actually using
tuning-rescale
|
| Mon, 16 Feb 2015 14:47:43 +0000 |
Chris Cannam |
Merge from refactors branch
tuning-rescale
|
| Fri, 13 Feb 2015 15:27:34 +0000 |
Chris Cannam |
Add parameters for min & max frequency
refactors
|
| Fri, 13 Feb 2015 15:18:47 +0000 |
Chris Cannam |
Add minimum and maximum frequency parameters to FeatureExtractor
refactors
|
| Fri, 06 Feb 2015 18:47:27 +0000 |
Chris Cannam |
Remove a bit of debug output
refactors
|
| Fri, 06 Feb 2015 18:41:44 +0000 |
Chris Cannam |
Disable debug
refactors
|
| Fri, 06 Feb 2015 18:41:35 +0000 |
Chris Cannam |
Merge from branch structure
refactors
|
| Fri, 06 Feb 2015 18:09:18 +0000 |
Chris Cannam |
Minor tidy
structure
|
| Fri, 06 Feb 2015 13:07:30 +0000 |
Chris Cannam |
Tidy up
structure
|
| Fri, 06 Feb 2015 11:46:23 +0000 |
Chris Cannam |
Fixes to copy ctor, operator= -- but these shouldn't be necessary, should restructure
structure
|
| Thu, 05 Feb 2015 16:26:41 +0000 |
Chris Cannam |
Alternative handling of reference frequency parameter -- scale the whole spectrum, don't just adjust the semitone bins above the crossover freq
tuning-rescale
|
| Thu, 05 Feb 2015 15:14:22 +0000 |
Chris Cannam |
Print queue length diagnostic
structure
|
| Thu, 05 Feb 2015 14:06:57 +0000 |
Chris Cannam |
Allow querying the best-estimate reference frame for the current feed point; don't heap-allocate finder
structure
|
| Thu, 05 Feb 2015 11:53:23 +0000 |
Chris Cannam |
Simplify freq handling
refactors
|
| Fri, 30 Jan 2015 14:54:18 +0000 |
Chris Cannam |
Build stuff
refactors
|
| Fri, 30 Jan 2015 12:57:29 +0000 |
Chris Cannam |
Add outputs & params to the ttl
refactors
|
| Thu, 29 Jan 2015 17:38:09 +0000 |
Chris Cannam |
Add overall cost output to plugin. Looks suspiciously unaffected by tuning frequency.
refactors
|
| Thu, 29 Jan 2015 17:22:02 +0000 |
Chris Cannam |
Tidy up parameter order &c
refactors
|
| Thu, 29 Jan 2015 17:13:58 +0000 |
Chris Cannam |
Expose tuning frequency parameters through plugin
refactors
|
| Thu, 29 Jan 2015 17:07:05 +0000 |
Chris Cannam |
Permit overriding reference frequency for second input in MatchPipeline
refactors
|
| Thu, 29 Jan 2015 17:02:48 +0000 |
Chris Cannam |
Introduce reference frequency param to feature extractor
refactors
|
| Thu, 29 Jan 2015 13:29:48 +0000 |
Chris Cannam |
Merge from refactors branch
structure
|
| Thu, 29 Jan 2015 10:55:24 +0000 |
Chris Cannam |
Default is actually Manhattan, not Euclidean (it just looks like squared-Euclidean for energy vectors). Add Euclidean as another alternative.
refactors
|
| Thu, 29 Jan 2015 10:25:47 +0000 |
Chris Cannam |
Implement distance metric selection (euclidean or cosine)
refactors
|
| Fri, 23 Jan 2015 17:33:37 +0000 |
Chris Cannam |
Make path retrieval available directly in pipeline; some ctor/operator= fixes (hopefully to become obsolete later)
structure
|
| Fri, 23 Jan 2015 14:55:19 +0000 |
Chris Cannam |
More cost query methods
structure
|
| Fri, 23 Jan 2015 11:46:14 +0000 |
Chris Cannam |
Merge
structure
|
| Fri, 23 Jan 2015 10:46:26 +0000 |
Chris Cannam |
Merge
structure
|
| Fri, 23 Jan 2015 09:20:04 +0000 |
Chris Cannam |
Make noise an option in the plugin: on by default, and therefore also switch on silence by default
refactors
|
| Thu, 22 Jan 2015 17:25:24 +0000 |
Chris Cannam |
Make noise a parameter in DistanceMetric (though not exposed in plugin etc yet)
refactors
|
| Thu, 22 Jan 2015 17:20:11 +0000 |
Chris Cannam |
Merge from branch "noise"
refactors
|
| Thu, 22 Jan 2015 16:55:27 +0000 |
Chris Cannam |
Some conversion fixes
structure
|
| Thu, 22 Jan 2015 12:04:44 +0000 |
Chris Cannam |
Factor out getBestEdgeCost
structure
|
| Fri, 16 Jan 2015 16:48:55 +0000 |
Chris Cannam |
Update unit tests
noise
|
| Fri, 16 Jan 2015 16:43:56 +0000 |
Chris Cannam |
Add a constant value (representing "noise") to each distance and sum
noise
|
| Fri, 16 Jan 2015 10:28:29 +0000 |
Chris Cannam |
Fix two errors introduced in merge
refactors
|
| Fri, 16 Jan 2015 10:18:00 +0000 |
Chris Cannam |
Transplant the distance metric parameter structure from silence_penalty branch (even though normalisation is currently the only thing in it)
refactors
|
| Thu, 15 Jan 2015 17:29:49 +0000 |
Chris Cannam |
Attempt naive silence penalty -- not currently working, but the plots are food for thought
silence_penalty
|
| Fri, 09 Jan 2015 17:20:52 +0000 |
Chris Cannam |
Merge from branch refactors
adaptive_diagonals
|
| Thu, 08 Jan 2015 12:11:27 +0000 |
Chris Cannam |
Separate out the raw & conditioned feature outputs (previously only conditioned was available, but we want raw for our tests). Plus some optional debug output
refactors
|
| Fri, 19 Dec 2014 15:07:57 +0000 |
Chris Cannam |
Fix uninitialised variable
refactors
|
| Fri, 19 Dec 2014 11:16:39 +0000 |
Chris Cannam |
Change defaults: no silence threshold, raw output by default
refactors
|
| Fri, 19 Dec 2014 10:30:09 +0000 |
Chris Cannam |
Bump plugin version
refactors
|
| Fri, 19 Dec 2014 10:19:33 +0000 |
Chris Cannam |
Expose silence threshold as parameter
refactors
|
| Thu, 18 Dec 2014 17:56:54 +0000 |
Chris Cannam |
The finder is supposed to use normalised path-cost when calculation expand direction (as in Java implementation). Also, provide a way to query the forward path.
refactors
|
| Thu, 11 Dec 2014 15:16:15 +0000 |
Chris Cannam |
Doc
refactors
|
| Thu, 11 Dec 2014 14:09:54 +0000 |
Chris Cannam |
Distance metric tests
refactors
|
| Thu, 11 Dec 2014 13:55:54 +0000 |
Chris Cannam |
Whoops, had checked in some executables mistaking them for headers in the list
refactors
|
| Thu, 11 Dec 2014 13:55:25 +0000 |
Chris Cannam |
Start DistanceMetric test
refactors
|
| Thu, 11 Dec 2014 13:53:16 +0000 |
Chris Cannam |
FeatureConditioner tests
refactors
|
| Thu, 11 Dec 2014 12:54:46 +0000 |
Chris Cannam |
Generalise chroma test at least to other frame/samplerate etc
refactors
|
| Thu, 11 Dec 2014 12:13:26 +0000 |
Chris Cannam |
Comment
refactors
|
| Thu, 11 Dec 2014 11:58:46 +0000 |
Chris Cannam |
Non-chroma test
refactors
|
| Thu, 11 Dec 2014 11:41:25 +0000 |
Chris Cannam |
Chroma feature test case
refactors
|
| Thu, 11 Dec 2014 11:41:11 +0000 |
Chris Cannam |
Docs
refactors
|
| Thu, 11 Dec 2014 10:16:26 +0000 |
Chris Cannam |
Prepare ground for some tests
refactors
|
| Thu, 11 Dec 2014 10:15:40 +0000 |
Chris Cannam |
Remove unused member
refactors
|
| Fri, 05 Dec 2014 18:06:51 +0000 |
Chris Cannam |
If two consecutive pin points have the same coordinate on one axis, drop the first. Also, don't forget to recalculate the opening and closing parts of the path
adaptive_diagonals
|
| Fri, 05 Dec 2014 17:28:53 +0000 |
Chris Cannam |
Move adaptive smoothing over to Finder::smooth
adaptive_diagonals
|
| Fri, 05 Dec 2014 15:19:48 +0000 |
Chris Cannam |
Use end point when recalculating with pin points
adaptive_diagonals
|
| Fri, 05 Dec 2014 14:49:57 +0000 |
Chris Cannam |
Confidence inflection points
adaptive_diagonals
|
| Fri, 05 Dec 2014 13:40:23 +0000 |
Chris Cannam |
Add raw features outputs, fix some path step bugs
adaptive_diagonals
|
| Fri, 05 Dec 2014 11:02:15 +0000 |
Chris Cannam |
Merge from refactors branch
adaptive_diagonals
|
| Fri, 05 Dec 2014 10:05:31 +0000 |
Chris Cannam |
Add OutputDerivative, for completeness, & make DistanceMetric cope with negative feature values
refactors
|
| Thu, 04 Dec 2014 17:18:39 +0000 |
Chris Cannam |
Some adjustments to intermediate features
adaptive_diagonals
|
| Thu, 04 Dec 2014 16:51:13 +0000 |
Chris Cannam |
Comment fix
refactors
|
| Thu, 04 Dec 2014 15:59:31 +0000 |
Chris Cannam |
Sample rate and fft size no longer needed in Matcher parameters
refactors
|
| Thu, 04 Dec 2014 15:55:05 +0000 |
Chris Cannam |
Merge from branch feature_conditioner
refactors
|
| Thu, 04 Dec 2014 15:44:05 +0000 |
Chris Cannam |
A feature hack (that is not going anywhere)
adaptive_diagonals
|
| Thu, 04 Dec 2014 15:24:27 +0000 |
Chris Cannam |
Obtain feature values from pipeline
adaptive_diagonals
|
| Thu, 04 Dec 2014 14:19:35 +0000 |
Chris Cannam |
Return distances through retrievePath
adaptive_diagonals
|
| Thu, 04 Dec 2014 13:54:05 +0000 |
Chris Cannam |
Merge from branch feature_conditioner, but leave the confidence calculation unfinished (need to move this)
adaptive_diagonals
|
| Thu, 04 Dec 2014 13:48:17 +0000 |
Chris Cannam |
Use MatchPipeline in MatchVampPlugin
feature_conditioner
|
| Thu, 04 Dec 2014 13:43:00 +0000 |
Chris Cannam |
Permit retrieving features
feature_conditioner
|
| Thu, 04 Dec 2014 13:35:23 +0000 |
Chris Cannam |
Start MatchPipeline
feature_conditioner
|
| Thu, 04 Dec 2014 13:23:10 +0000 |
Chris Cannam |
Feature size is not actually needed in Matcher
feature_conditioner
|
| Thu, 04 Dec 2014 13:05:16 +0000 |
Chris Cannam |
Pull out normalisation and specdiff stuff into FeatureConditioner
feature_conditioner
|
| Thu, 04 Dec 2014 10:24:35 +0000 |
Chris Cannam |
Merge
adaptive_diagonals
|
| Thu, 04 Dec 2014 08:49:45 +0000 |
Chris Cannam |
Import median filter
adaptive_diagonals
|
| Mon, 01 Dec 2014 17:54:52 +0000 |
Chris Cannam |
Test implementations of confidence peaks & adaptive smoothing
adaptive_diagonals
|
| Fri, 28 Nov 2014 14:47:16 +0000 |
Chris Cannam |
Make the magnitude feature be the sum of magnitudes, not just magnitude of first file
adaptive_diagonals
|
| Thu, 27 Nov 2014 18:01:36 +0000 |
Chris Cannam |
Add feature distance and feature magnitude outputs (for experiments)
adaptive_diagonals
|
| Thu, 27 Nov 2014 17:08:22 +0000 |
Chris Cannam |
Close cheap_diagonals branch, now that diagonal weight is exposed as a parameter in the refactors branch
cheap_diagonals
|
| Thu, 27 Nov 2014 16:50:14 +0000 |
Chris Cannam |
Tidy
refactors
|
| Thu, 27 Nov 2014 16:05:24 +0000 |
Chris Cannam |
More debug output
refactors
|
| Thu, 27 Nov 2014 16:05:19 +0000 |
Chris Cannam |
Ah, it's subtler than that -- we have to handle the case where this is not the first matcher as well, if we're to get precisely the same results
refactors
|
| Thu, 27 Nov 2014 15:56:12 +0000 |
Chris Cannam |
Fix inconsistency (with Finder) in choice of preference where two costs come out equal
refactors
|
| Thu, 27 Nov 2014 15:29:43 +0000 |
Chris Cannam |
Improve error reporting
refactors
|
| Thu, 27 Nov 2014 13:02:28 +0000 |
Chris Cannam |
Remove some debug out
refactors
|
| Thu, 27 Nov 2014 12:59:28 +0000 |
Chris Cannam |
And remove that printout
refactors
|
| Thu, 27 Nov 2014 12:59:23 +0000 |
Chris Cannam |
Fix diagonal handling: there was an implicit fixed diagonal weight of 2 in comparison of diagmin+distance to othermin directly
refactors
|
| Thu, 27 Nov 2014 12:57:59 +0000 |
Chris Cannam |
Version which prints out matrix on error
refactors
|
| Thu, 27 Nov 2014 12:54:20 +0000 |
Chris Cannam |
Move some functions about
refactors
|
| Thu, 27 Nov 2014 12:08:16 +0000 |
Chris Cannam |
Previous commit was a mistake: the ahead-of-time business is in Finder::getExpandDirection. In fact we were failing to swap advance directions in forward path when writing to the "other" finder. This does not actually affect the backward path calculation, but it does mean we can restore the sanity check.
refactors
|
| Thu, 27 Nov 2014 11:27:17 +0000 |
Chris Cannam |
Ah, no, of course the advance differs for the forward path (it is based on partial information). But the cost shouldn't.
refactors
|
| Thu, 27 Nov 2014 10:53:00 +0000 |
Chris Cannam |
More error checking (check advance as well as cost)
refactors
|
| Thu, 27 Nov 2014 10:49:11 +0000 |
Chris Cannam |
Expose diagonal weight as a parameter
refactors
|
