| Thu, 09 Apr 2020 14:59:05 +0100 |
Chris Cannam |
Fix out of range access to magnitudes
|
| Thu, 09 Apr 2020 11:22:55 +0100 |
Chris Cannam |
Unify various vectors to our base floatvec_t type; store columns in fft model cache at their desired height so we can return a reference (speeding up the peak-frequency spectrogram in particular)
|
| Wed, 25 Sep 2019 09:43:34 +0100 |
Chris Cannam |
Some tidying
time-frequency-boxes
|
| Tue, 17 Sep 2019 11:21:33 +0100 |
Chris Cannam |
More graceful handling of failure to construct FFT models in the case where the source model has already been deleted before this occurs
|
| Thu, 12 Sep 2019 11:52:19 +0100 |
Chris Cannam |
Support optional max frequency setting, useful when we want to store caches of very constrained frequency ranges (as in melodic-range spectrogram, potentially)
|
| Wed, 14 Aug 2019 13:54:23 +0100 |
Chris Cannam |
This output is too annoying, in the perfectly innocuous case of reading from an aggregate model whose components are different lengths
|
| Fri, 05 Jul 2019 15:28:07 +0100 |
Chris Cannam |
Further review of ById: make IDs only available when adding a model to the ById store, not by querying the item directly. This means any id encountered in the wild must have been added to the store at some point (even if later released), which simplifies reasoning about lifecycles
by-id
|
| Tue, 02 Jul 2019 11:49:28 +0100 |
Chris Cannam |
Update FFT model to ById
by-id
|
| Wed, 24 Apr 2019 11:44:32 +0100 |
Chris Cannam |
Merge from default branch
single-point
|
| Thu, 04 Apr 2019 11:15:43 +0100 |
Chris Cannam |
Model deletion issues - the AlignmentModel doesn't actually need to know about the input aggregate model; the document should own it instead
|
| Thu, 04 Apr 2019 11:15:43 +0100 |
Chris Cannam |
Model deletion issues - the AlignmentModel doesn't actually need to know about the input aggregate model; the document should own it instead
single-point
|
| Fri, 29 Mar 2019 19:16:11 +0000 |
Chris Cannam |
Fix error message which says the opposite of the truth
single-point
|
| Mon, 26 Nov 2018 14:33:41 +0000 |
Chris Cannam |
Use nullptr throughout
fix-static-analysis
|
| Tue, 13 Nov 2018 13:29:37 +0000 |
Chris Cannam |
Further adjustments to peak picking, to try to avoid a surfeit of peaks in the higher frequencies
spectrogramparam
|
| Mon, 12 Nov 2018 14:48:15 +0000 |
Chris Cannam |
Slightly simplify median-window size calculation - but this is not the best way to do this for spectrum views, need to reconsider
spectrogramparam
|
| Thu, 08 Nov 2018 15:02:30 +0000 |
Chris Cannam |
Introduce & make use of faster MovingMedian class (now with resize capability)
spectrogramparam
|
| Thu, 08 Nov 2018 12:56:03 +0000 |
Chris Cannam |
A little faster as vector
spectrogramparam
|
| Wed, 07 Nov 2018 15:46:20 +0000 |
Chris Cannam |
Comment
|
| Wed, 07 Nov 2018 11:47:24 +0000 |
Chris Cannam |
Properly align window with signal, in oversampled case
|
| Tue, 24 Apr 2018 14:13:51 +0100 |
Chris Cannam |
Some messing around with profiling
|
| Thu, 01 Mar 2018 18:02:22 +0000 |
Chris Cannam |
Untabify
|
| Thu, 01 Mar 2018 14:43:40 +0000 |
Chris Cannam |
Convert some cerrs to SVCERRs. Apart from anything else, this makes MSVC2017 happy
|
| Thu, 26 Jan 2017 14:11:20 +0000 |
Chris Cannam |
Reduce amount of copying in FFTModel
|
| Tue, 13 Dec 2016 15:23:21 +0000 |
Chris Cannam |
Update to use bqvec allocator for float vectors
3.0-integration
|
| Wed, 07 Dec 2016 13:56:22 +0000 |
Chris Cannam |
Remove unnecessary move()s
3.0-integration
|
| Wed, 23 Nov 2016 10:34:45 +0000 |
Chris Cannam |
Debug note when we overrun end of data
3.0-integration
|
| Mon, 21 Nov 2016 15:05:57 +0000 |
Chris Cannam |
Branch to use bqfft code and remove FFTapi
bqfft
|
| Thu, 10 Nov 2016 09:58:28 +0000 |
Chris Cannam |
More timings and cache hit counts
3.0-integration
|
| Thu, 10 Nov 2016 09:23:05 +0000 |
Chris Cannam |
A note on cache hit findings
3.0-integration
|
| Wed, 09 Nov 2016 18:08:40 +0000 |
Chris Cannam |
Print out cache hit/miss counts
3.0-integration
|
