| Fri, 07 Feb 2020 11:49:39 +0000 |
Chris Cannam |
Ensure default parameter values match the actual internal defaults
spect tip
|
| Thu, 04 Oct 2018 16:06:29 +0100 |
Chris Cannam |
Remove inaccurate comment
spect
|
| Thu, 04 Oct 2018 16:00:50 +0100 |
Chris Cannam |
Avoid compiler warnings
spect
|
| Thu, 04 Oct 2018 15:42:03 +0100 |
Chris Cannam |
Experiment at limiting number of partials (for speed). Interacts badly with harmonic correction, so doesn't work at the mo
maxpartial
|
| Thu, 04 Oct 2018 15:22:23 +0100 |
Chris Cannam |
Remove nfft for unimplemented oversampling option; rejig nsamp param to use indexed list
spect
|
| Thu, 04 Oct 2018 14:57:21 +0100 |
Chris Cannam |
Remove a parameter that never had any implementation
spect
|
| Thu, 04 Oct 2018 14:56:41 +0100 |
Chris Cannam |
Connect up an interesting parameter that was exposed but not connected
spect
|
| Thu, 04 Oct 2018 14:46:41 +0100 |
Chris Cannam |
Some tidying
spect
|
| Thu, 04 Oct 2018 13:32:47 +0100 |
Chris Cannam |
Correct the time-alignment of the output blocks
spect
|
| Thu, 04 Oct 2018 11:38:47 +0100 |
Chris Cannam |
Add a simple most-salient-pitch output (making a pitch tracker)
spect
|
| Thu, 04 Oct 2018 10:17:32 +0100 |
Chris Cannam |
Change calculation of frequency index based on the usual FFT bin frequencies. I'm troubled by this change, but it does appear to produce the correct result.
spect
|
| Thu, 04 Oct 2018 09:46:30 +0100 |
Chris Cannam |
Avoid compiler warnings
spect
|
| Wed, 03 Oct 2018 15:47:00 +0100 |
Chris Cannam |
Toward (but not quite reaching) accurate frequency labels for outputs
spect
|
| Wed, 03 Oct 2018 13:16:09 +0100 |
Chris Cannam |
Break out the plugin into three different plugins (using the same class in different modes) in order to provide simplistic and more sophisticated spectrograms as well as the f0-gram. Remove the program support, since it doesn't work usefully anyway (it just overrides the user's preferred settings).
spect
|
| Tue, 02 Oct 2018 16:38:52 +0100 |
Chris Cannam |
Switch to Vamp SDK FFT implementation (it is close enough in performance - FFTs aren't really a bottleneck here - and simpler for the build) and use bqvec allocators
perf
|
| Tue, 02 Oct 2018 13:49:23 +0100 |
Chris Cannam |
Remove further unused
perf
|
| Tue, 02 Oct 2018 13:21:15 +0100 |
Chris Cannam |
size_t -> int throughout: permits better optimisation for tight for-loops, making the whole thing about 10% faster
perf
|
| Tue, 02 Oct 2018 13:17:19 +0100 |
Chris Cannam |
Remove some commented-out code
perf
|
| Tue, 02 Oct 2018 13:14:21 +0100 |
Chris Cannam |
Indent
perf
|
| Tue, 02 Oct 2018 13:13:36 +0100 |
Chris Cannam |
Remove some debug
perf
|
| Tue, 02 Oct 2018 13:07:51 +0100 |
Chris Cannam |
Avoid warning
|
| Tue, 02 Oct 2018 10:44:42 +0100 |
Chris Cannam |
Import original code from the downloaded VampFChTCore-v1.1beta archive
|
