| Wed, 07 Oct 2015 16:47:31 +0100 |
Chris Cannam |
Fix crash in finetune mode sometimes caused by note remaining on while column falls below threshold
|
| Mon, 07 Sep 2015 16:23:55 +0100 |
Chris Cannam |
Copy the column before potentially erasing while iterating through it
|
| Mon, 07 Sep 2015 14:45:19 +0100 |
Chris Cannam |
Some tidying, switch to libc++
|
| Thu, 03 Sep 2015 12:06:31 +0100 |
Chris Cannam |
Restore threads stuff -- the host needs to be compiled with threading
|
| Wed, 12 Aug 2015 17:27:51 +0100 |
Chris Cannam |
Be sure to emit offsets even after very short note-part
mirex2015
|
| Wed, 12 Aug 2015 16:03:53 +0100 |
Chris Cannam |
Annoyingly, getting crashes in the tester when using C++11 futures -- for now simply disable that, but it'll need looking into
|
| Wed, 12 Aug 2015 15:01:19 +0100 |
Chris Cannam |
Fix incorrect max value for processing mode parameter
|
| Wed, 12 Aug 2015 14:50:30 +0100 |
Chris Cannam |
Make the onset/offset output consistent with the notes output in terms of restarting notes when a shift changes
|
| Mon, 06 Jul 2015 09:15:09 +0100 |
Chris Cannam |
Replace Draft with Live mode
livemode
|
| Sat, 27 Jun 2015 14:37:13 +0100 |
Chris Cannam |
An abortive attempt at this (doesn't compile)
simultaneities
|
| Fri, 26 Jun 2015 10:23:54 +0100 |
Chris Cannam |
Add onset+offset output; look up shift counts rather than passing them around
livemode
|
| Thu, 25 Jun 2015 17:23:31 +0100 |
Chris Cannam |
Back out this reduction of duration threshold, I think it makes things too noisy
livemode
|
| Thu, 25 Jun 2015 14:18:44 +0100 |
Chris Cannam |
Detect repeated notes
livemode
|
| Thu, 25 Jun 2015 12:21:02 +0100 |
Chris Cannam |
Merge
livemode
|
| Wed, 29 Apr 2015 18:58:00 +0100 |
Chris Cannam |
Merge
livemode
|
| Tue, 28 Apr 2015 18:56:54 +0100 |
Chris Cannam |
Merge
livemode
|
| Tue, 28 Apr 2015 18:56:34 +0100 |
Chris Cannam |
Minor text fix
livemode
|
| Tue, 19 May 2015 09:29:00 +0100 |
Chris Cannam |
Experiment with dropping the bottom octave off each template (since most of the information is in higher harmonics anyway!) -- this is about 15% faster again and has half the latency, but per
livemode-octave-higher
|
| Mon, 18 May 2015 13:58:27 +0100 |
Chris Cannam |
Always use q=0.8 and accept the hit on speed -- the templates are made for that configuration and it does work better. Also some adjustments to thresholding and peak picking for live mode in particular.
livemode
|
| Wed, 29 Apr 2015 10:56:30 +0100 |
Chris Cannam |
Fix oversight which saw note tracking use pre-filtered pitches (resulting in many short notes)
livemode
|
| Wed, 29 Apr 2015 10:12:02 +0100 |
Chris Cannam |
Shorten duration threshold for live mode; return non-peak-picked pitch distribution from pitch distribution output
livemode
|
| Wed, 29 Apr 2015 09:51:09 +0100 |
Chris Cannam |
Refactor velocity calculations
livemode
|
| Wed, 29 Apr 2015 09:27:38 +0100 |
Chris Cannam |
Add onsets output (lower-latency than the notes output)
livemode
|
| Tue, 28 Apr 2015 13:55:32 +0100 |
Chris Cannam |
Some small speed improvements for live mode (+ don't use async for it when there's only one template in the EM process: the overhead isn't worth it)
livemode
|
| Tue, 28 Apr 2015 12:21:40 +0100 |
Chris Cannam |
Live mode enhancements: Adjust processing parameters (for speed) and peak-pick pitch activations across frequency (to avoid neighbouring-semitone clusters)
livemode
|
| Tue, 28 Apr 2015 11:24:23 +0100 |
Chris Cannam |
Merge from default branch
livemode
|
| Tue, 28 Apr 2015 11:09:31 +0100 |
Chris Cannam |
Don't rely on column count to determine whether we've set the start time or not -- it could theoretically be 0 even after some input has appeared
|
| Tue, 28 Apr 2015 10:02:36 +0100 |
Chris Cannam |
Replace OpenMP usage with (hopefully more portable) C++11 async. Define MAX_EM_THREADS=1 if you don't want to use it.
|
| Tue, 28 Apr 2015 09:41:40 +0100 |
Chris Cannam |
Pull out applyEM function
|
| Mon, 19 Jan 2015 11:23:07 +0000 |
Chris Cannam |
Add chromagram output
|
| Fri, 05 Dec 2014 17:00:06 +0000 |
Chris Cannam |
Adjust silent column threshold for live mode (fewer bins)
livemode
|
| Fri, 05 Dec 2014 16:47:06 +0000 |
Chris Cannam |
Merge from default branch
livemode
|
| Fri, 05 Dec 2014 16:46:57 +0000 |
Chris Cannam |
Now that we are returning the pitch activation matrix, we don't want to skip silent columns in the postprocess loop
|
| Fri, 05 Dec 2014 16:40:51 +0000 |
Chris Cannam |
Bump version for live mode
livemode
|
| Fri, 05 Dec 2014 16:34:24 +0000 |
Chris Cannam |
More adjustments for live mode, including actually *using* the max frequency we calculated earlier
livemode
|
| Tue, 02 Dec 2014 17:13:10 +0000 |
Chris Cannam |
Add output for the templates (probably temporarily)
livemode
|
| Mon, 01 Dec 2014 17:12:19 +0000 |
Chris Cannam |
More tweaking of frequencies
livemode
|
| Fri, 28 Nov 2014 18:15:44 +0000 |
Chris Cannam |
A thresholding "fix"
livemode
|
| Fri, 28 Nov 2014 10:18:22 +0000 |
Chris Cannam |
More toward a possible live mode
livemode
|
| Fri, 28 Nov 2014 09:42:56 +0000 |
Chris Cannam |
Thinking about a "live" mode...
livemode
|
| Fri, 17 Oct 2014 09:26:59 +0100 |
Chris Cannam |
Bump plugin version
|
| Wed, 15 Oct 2014 17:40:20 +0100 |
Chris Cannam |
Add pitch activation matrix output
|
| Thu, 07 Aug 2014 16:06:02 +0100 |
Chris Cannam |
Fix failure to take starting timestamp into account
|
| Mon, 28 Jul 2014 14:40:44 +0100 |
Chris Cannam |
Handle resampler latency, impose limits on input sample rate
|
| Sat, 26 Jul 2014 10:18:01 +0100 |
Chris Cannam |
Docs, fix to temporary file removal in test script
|
| Wed, 23 Jul 2014 10:20:07 +0100 |
Chris Cannam |
Make use of input gain information to adjust output note velocities back to original levels
norm
|
| Tue, 22 Jul 2014 16:36:21 +0100 |
Chris Cannam |
Rename some output files; test flattendynamics-using plugin with un-normalised TRIOS input
norm
|
| Tue, 22 Jul 2014 16:11:20 +0100 |
Chris Cannam |
Use flattendynamics code within Silvet
norm eval-norm-u4
|
| Thu, 17 Jul 2014 16:42:48 +0100 |
Chris Cannam |
Try using the Flatten Dynamics plugin instead of norm-to-max-so-far
norm eval-norm-rN
|
| Tue, 15 Jul 2014 13:32:27 +0100 |
Chris Cannam |
Normalise samples to cumulative maximum (experimentally)
norm
|
| Tue, 15 Jul 2014 13:28:25 +0100 |
Chris Cannam |
Make source sparsity configurable in instrument pack as well as pitch sparsity; make both equal to 1 for piano templates
|
| Wed, 04 Jun 2014 11:46:01 +0100 |
Chris Cannam |
Merge from the default branch
noteagent
|
| Wed, 04 Jun 2014 09:55:31 +0100 |
Chris Cannam |
RDF file, and some copyright etc updates
|
| Thu, 29 May 2014 10:30:08 +0100 |
Chris Cannam |
Fixes to mono feeder; use it for monophonic instruments
noteagent
|
| Thu, 29 May 2014 09:21:15 +0100 |
Chris Cannam |
Emit "MIDI-compatible" frequencies only, unless in fine tuning mode
noteagent
|
| Wed, 28 May 2014 15:16:00 +0100 |
Chris Cannam |
Don't retain returned hypotheses -- to avoid having to test whether we've seen them before
noteagent
|
| Wed, 28 May 2014 15:05:43 +0100 |
Chris Cannam |
Tidy column counting
noteagent
|
| Wed, 28 May 2014 14:56:01 +0100 |
Chris Cannam |
Merge from default branch
noteagent
|
| Wed, 28 May 2014 14:54:01 +0100 |
Chris Cannam |
Wire up note agent code -- results are not very good, so far
noteagent
|
| Fri, 23 May 2014 18:17:59 +0100 |
Chris Cannam |
Add some further parameters to the instrument pack
|
| Thu, 22 May 2014 16:59:12 +0100 |
Chris Cannam |
Have a first go at reporting f0
f0
|
| Thu, 22 May 2014 15:06:37 +0100 |
Chris Cannam |
Fix incorrect use of int for non-integer shift weights
|
| Thu, 22 May 2014 12:11:33 +0100 |
Chris Cannam |
Restore (and fix) the time-frequency distribution output (filtered CQ); tidy up the bottom-octave handling in draft mode
|
| Wed, 21 May 2014 15:03:12 +0100 |
Chris Cannam |
Add some ensemble instrument packs
|
| Wed, 21 May 2014 14:42:00 +0100 |
Chris Cannam |
Emit cents offset in note name too
|
| Wed, 21 May 2014 12:15:29 +0100 |
Chris Cannam |
In draft mode, make the CQ transform one octave shorter (bottom octave empty)
|
| Wed, 21 May 2014 12:10:02 +0100 |
Chris Cannam |
Remove comment (quick test suggests this makes little difference either way)
|
| Wed, 21 May 2014 11:53:31 +0100 |
Chris Cannam |
Refactor to avoid allocating 300 EM objects at a time!
finetune
|
| Wed, 21 May 2014 11:15:53 +0100 |
Chris Cannam |
More work on returning sensibly-shifted notes with fine tuning
finetune
|
| Wed, 21 May 2014 09:11:49 +0100 |
Chris Cannam |
Store active shifts separately from active pitches
finetune
|
| Tue, 20 May 2014 17:49:07 +0100 |
Chris Cannam |
More on returning fine tuning (but we're treating different shifts of the same pitch as different notes at the moment which is not right)
finetune
|
| Tue, 20 May 2014 17:34:49 +0100 |
Chris Cannam |
Start working on returning fine-tuning pitches
finetune
|
| Tue, 20 May 2014 16:48:33 +0100 |
Chris Cannam |
Use 20ms resolution in hq mode, 40ms only in draft mode
|
| Tue, 20 May 2014 16:21:33 +0100 |
Chris Cannam |
Some measurements suggest we get slightly better results with more iterations, but not drastically worse with fewer -- so go lower for draft and higher for hq mode
|
| Tue, 20 May 2014 15:56:57 +0100 |
Chris Cannam |
Minor tidy
|
| Tue, 20 May 2014 15:37:03 +0100 |
Chris Cannam |
Add solo instrument parameter
|
| Tue, 20 May 2014 15:31:05 +0100 |
Chris Cannam |
Introduce InstrumentPack class
|
| Tue, 20 May 2014 13:55:44 +0100 |
Chris Cannam |
Remove extra outputs
|
| Mon, 19 May 2014 17:47:34 +0100 |
Chris Cannam |
Bin frequencies, and some comments
|
| Fri, 16 May 2014 10:28:46 +0100 |
Chris Cannam |
Note on parameters etc
|
| Fri, 16 May 2014 10:12:55 +0100 |
Chris Cannam |
Processing parameters. Main difference is use of atomHopFactor of 0.3 instead of 0.25
|
| Thu, 15 May 2014 15:25:03 +0100 |
Chris Cannam |
Update subrepo to new structure
|
| Wed, 14 May 2014 19:38:36 +0100 |
Chris Cannam |
double -> float in EM (to test)
|
| Wed, 14 May 2014 18:09:06 +0100 |
Chris Cannam |
Some comments on things to try next...
|
| Tue, 13 May 2014 14:34:04 +0100 |
Chris Cannam |
Update CQ implementation following inverse-CQ merge
|
| Wed, 07 May 2014 11:06:18 +0100 |
Chris Cannam |
Simplify omp loop, leaving more to the omp scheduler
bqvec-openmp
|
| Wed, 07 May 2014 09:43:15 +0100 |
Chris Cannam |
Pull out shifted templates into separate arrays with proper alignment (trading off memory occupied against getting the alignments right)
bqvec-openmp
|
| Wed, 07 May 2014 09:08:52 +0100 |
Chris Cannam |
Simplify slightly, make HQ mode the default
bqvec-openmp
|
| Wed, 07 May 2014 09:01:32 +0100 |
Chris Cannam |
Merge from default branch
bqvec-openmp
|
| Tue, 06 May 2014 18:55:11 +0100 |
Chris Cannam |
Add draft/intensive mode setting (determines whether to use shifts)
|
| Tue, 06 May 2014 16:36:46 +0100 |
Chris Cannam |
Merge OpenMP and bqvec stuff into bqvec-openmp branch
bqvec-openmp
|
| Tue, 06 May 2014 16:28:04 +0100 |
Chris Cannam |
OpenMP support in main EM iteration
openmp
|
| Tue, 06 May 2014 14:45:16 +0100 |
Chris Cannam |
Use single-precision floats throughout EM code
timing
|
| Tue, 06 May 2014 14:19:19 +0100 |
Chris Cannam |
Complete using bqvec for allocation etc, but with unchanged processing logic
bqvec
|
| Tue, 06 May 2014 11:26:33 +0100 |
Chris Cannam |
Timing outputs
timing
|
| Tue, 06 May 2014 11:23:07 +0100 |
Chris Cannam |
Halve number of iterations
timing
|
| Tue, 06 May 2014 11:21:28 +0100 |
Chris Cannam |
Some testing results
timing
|
| Tue, 06 May 2014 11:17:08 +0100 |
Chris Cannam |
Remove debug printouts
timing
|
| Tue, 06 May 2014 11:13:25 +0100 |
Chris Cannam |
Remove FCQ return as well
timing
|
| Tue, 06 May 2014 11:10:52 +0100 |
Chris Cannam |
Remove return of CQ output features
timing
|
| Tue, 29 Apr 2014 12:05:52 +0100 |
Chris Cannam |
Run again, with files normalised as in the MIREX tests
|
| Mon, 28 Apr 2014 13:01:30 +0100 |
Chris Cannam |
Test code
|
| Tue, 08 Apr 2014 17:45:26 +0100 |
Chris Cannam |
Comment
|
| Tue, 08 Apr 2014 13:36:07 +0100 |
Chris Cannam |
Max, not median strength for velocity (duh)
|
| Tue, 08 Apr 2014 13:30:32 +0100 |
Chris Cannam |
Support shifts as an additional dimension (as in the original model). Also return velocity as well.
preshift
|
| Mon, 07 Apr 2014 17:07:51 +0100 |
Chris Cannam |
Merge from default branch
preshift
|
| Mon, 07 Apr 2014 17:07:36 +0100 |
Chris Cannam |
More outputs
|
| Mon, 07 Apr 2014 16:23:16 +0100 |
Chris Cannam |
Merge from default branch
preshift
|
| Mon, 07 Apr 2014 16:21:57 +0100 |
Chris Cannam |
Use supplied pitch ranges, plus subrepo update
|
| Mon, 07 Apr 2014 14:36:52 +0100 |
Chris Cannam |
Sum rather than max?
preshift
|
| Mon, 07 Apr 2014 14:18:39 +0100 |
Chris Cannam |
Merge from default branch
preshift
|
| Mon, 07 Apr 2014 14:18:22 +0100 |
Chris Cannam |
Fix noise-reduction filtering (it was going the wrong way)
|
| Mon, 07 Apr 2014 13:01:08 +0100 |
Chris Cannam |
Get notes out using pre-shifted templates (not working properly)
preshift
|
| Sat, 05 Apr 2014 17:57:50 +0100 |
Chris Cannam |
Return an actual note transcription
|
| Sat, 05 Apr 2014 13:18:55 +0100 |
Chris Cannam |
Return a sketch of notes from pitch activation matrix
|
| Sat, 05 Apr 2014 08:41:55 +0100 |
Chris Cannam |
Fix typo in pitch activation output
|
| Fri, 04 Apr 2014 19:05:47 +0100 |
Chris Cannam |
Return pitch activation matrix
|
| Fri, 04 Apr 2014 18:18:18 +0100 |
Chris Cannam |
Calculate column sums, etc
|
| Fri, 04 Apr 2014 17:48:06 +0100 |
Chris Cannam |
Fill out the EM logic
|
| Fri, 04 Apr 2014 14:28:41 +0100 |
Chris Cannam |
Start on EM class
|
