david/librubberband
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
426 Commits 3 Branches 0 Tags
05db791db3f8d27d7959957b407f228158cafc74
Commit Graph

187 Commits

Author SHA1 Message Date
Chris Cannam
674f44dcba Minor tidy 2021-10-14 10:03:37 +01:00
Chris Cannam
e868ff32de Remove some hopefully unnecessary references to global namespace 2021-10-12 16:27:19 +01:00
Chris Cannam
1f8173b619 Further relative paths 2021-10-04 12:29:48 +01:00
Chris Cannam
61f7b5f8e6 Update includes so they don't require include path to be set 2021-10-04 12:21:39 +01:00
Chris Cannam
66354796b8 Pull in multi-channel resampler fix (not directly relevant to RB) 2021-09-30 15:36:21 +01:00
Chris Cannam
c8087b7d43 Merge from default branch 2021-08-23 11:57:56 +01:00
Chris Cannam
ad27ad1b24 Fix failure to pack Nyquist bins correctly in vDSP FFT - though these functions are not actually used by Rubber Band, they're just here for symmetry 2021-07-05 19:28:05 +01:00
Chris Cannam
b199400e8b Fix failure to resize buffer on reconfigure - thanks to Michael Tyson 2021-06-01 08:49:54 +01:00
Chris Cannam
d22cbf3098 Comments 2021-05-14 15:25:54 +01:00
Chris Cannam
27a2980a56 Restore original behaviour of pre-fill/start-skip 2021-05-14 15:10:24 +01:00
Chris Cannam
a218ad42fa Constrain to +/- 2 octaves (+ semitones) as it's less demanding of block size; minor tidying 2021-05-14 14:26:52 +01:00
Chris Cannam
a3aff961e4 Avoid identifying transients when the divergence is particularly high already, so as not to pile on further 2021-05-14 14:24:54 +01:00
Chris Cannam
d9dfc0a84c Start updating LADSPA plugin for proper timing 2021-05-14 11:35:08 +01:00
Chris Cannam
478d9233fe Fix calculation of recovery value now that divergence is recalculated afresh on each update 2021-05-14 08:46:37 +01:00
Chris Cannam
d06b4efc16 Toward more accurate timing in the face of varying pitch ratio 2021-05-13 18:04:43 +01:00
Chris Cannam
6aeb713921 Use proper parameters for processing mode 2021-05-12 17:29:26 +01:00
Chris Cannam
c1fd6fe6a5 Fix overrun 2021-05-12 17:26:27 +01:00
Chris Cannam
94d4467bfe Fix tsan reports. This does mean updating to C++11 2021-05-12 11:12:02 +01:00
Chris Cannam
b7770c8832 Use the appropriate parameters 2021-05-11 17:20:12 +01:00
Chris Cannam
f6a66171bc Initial experimental import of bq resampler 2021-05-10 18:11:35 +01:00
Chris Cannam
e1e2338f23 Merge from default branch 2021-05-10 08:56:30 +01:00
Chris Cannam
cc937ebe65 Avoid a "using namespace" 2021-04-14 19:14:32 +01:00
Chris Cannam
e7f47b4b98 Introduce built-in fft; drop no-longer-supported ffts 2021-04-08 14:23:27 +01:00
Chris Cannam
91274b3fd5 Build and install JNI if header and Java compiler are found 2021-03-12 09:52:54 +00:00
Chris Cannam
a5cb3a6b4a Add some profiler entries 2021-03-11 16:15:10 +00:00
Chris Cannam
09e5bd29bf Protect profiler map with a mutex 2021-03-11 16:14:40 +00:00
Chris Cannam
fcc952e81a A few helper functions 2021-03-09 17:29:21 +00:00
Chris Cannam
b463cf5565 Fix wrong accumulator type - this fix makes the detector a little more sensitive, which is not necessarily good, but the previous accidental behaviour exaggerated the difference in results between float and double precision ffts 2021-03-09 17:22:41 +00:00
Chris Cannam
e79658b04f Fix incorrect cases in double-only/float-only cases 2021-03-09 17:21:08 +00:00
Chris Cannam
ee9128020d Additional kissfft source file 2021-03-09 13:05:24 +00:00
Chris Cannam
b81d62bb98 Update KissFFT code 2021-03-09 12:29:40 +00:00
Chris Cannam
120edca223 Build updates for Windows 2021-03-05 10:31:00 +00:00
Chris Cannam
7c8fa72c60 Avoid reserved preprocessor symbols 2021-02-10 11:07:33 +00:00
Chris Cannam
d2e5cf3a6d Fix incorrect initialisation 2021-02-05 16:39:41 +00:00
Chris Cannam
955a32fc7e Copyright dates 2021-01-08 17:13:52 +00:00
Chris Cannam
16487324fc Replace __MSVC__ define (which needs to be defined manually in the build) with _MSC_VER (which the compiler defines itself) throughout 2020-12-10 09:25:28 +00:00
Chris Cannam
897bd14b8e Avoid any smoothing when setting initial ratio 2020-11-16 09:11:12 +00:00
Chris Cannam
dbaaf12598 Hm, those settings weren't as well-behaved as I thought. I wonder if there is some unpleasant randomness here 2020-10-22 17:13:55 +01:00
Chris Cannam
e3e5fe7ae3 Settle on a single set of params that seems to work ok in tests 2020-10-22 15:55:48 +01:00
Chris Cannam
75f9af1f9b Remove support for resampler from IPP versions older than v7 2020-10-22 14:00:32 +01:00
Chris Cannam
cce7bdd703 Fix buffer underrun in case where rounded sample count accumulator exceeds length 2020-10-22 13:55:50 +01:00
Chris Cannam
c4ad5b6f4a Also avoid passing a genuinely long block when the ratio has just changed 2020-10-21 17:16:15 +01:00
Chris Cannam
ce5b79bb45 Avoid passing a very oversize outcount for the reasons given 2020-10-21 16:50:19 +01:00
Chris Cannam
3ddc35f7ae Resampler code pulled back in from bqresample, with safer api 2020-10-21 15:05:14 +01:00
Chris Cannam
81629cf714 Merge from default branch 2020-10-21 09:07:44 +01:00
Chris Cannam
d05806450a Avoid compiler complaints about dynamic array syntax. This is no better code, bu 
t it's less sensitive to compile options.
 2020-10-08 11:20:52 +01:00
Chris Cannam
42ff18a229 Prefer libsamplerate at any performance setting, now we know it is better-behaved when changing rates 2020-09-29 11:04:18 +01:00
Chris Cannam
c1108834c1 In this emergency, increase the buffer size by a factor rather than only by a constant amount - because if this arises because of an overlong increment being broken down into littler bits, we would end up doing it again and again, which is much more expensive. This makes a huge performance difference in cases where a time map ends up mapping very little input into a great expanse of output 2020-09-22 10:35:50 +01:00
Chris Cannam
8df9558868 In this emergency, increase the buffer size by a factor rather than only by a constant amount - because if this arises because of an overlong increment being broken down into littler bits, we would end up doing it again and again, which is much more expensive. This makes a huge performance difference in cases where a time map ends up mapping very little input into a great expanse of output 2020-09-22 10:35:50 +01:00
Chris Cannam
f42a369b8e Same fix in resampleInterleaved as earlier committed in resample 2020-09-21 16:44:53 +01:00