7e03d2bd3d
Use naopt NeuralAudio branch
2025-02-05 10:40:59 -08:00
368e47e12e
Update NeuralAudio (better RTNeural NAM performance)
2024-11-27 09:17:49 -08:00
9d2e7ae205
Fixed propagation of native arch optimizations
2024-11-16 08:09:58 -08:00
eeaa4649a4
Update NeuralAudio (add ".aidax" extension")
2024-11-15 07:25:59 -08:00
3f5872cc78
CMake options for forcing use of NAM Core for LSTM or WaveNet
2024-11-14 07:47:52 -08:00
32f18a242a
Handle model input gain adjustment
2024-11-04 07:35:34 -08:00
19d3abb12d
Switch to NeuralAudio for model handling
2024-10-31 10:35:28 -07:00
4643e66e08
Removed dc blocker for now
2024-10-31 08:29:42 -07:00
61bbce1e2a
Add "nam" namespace
2024-08-17 11:23:52 -07:00
72fd4a1e3d
Make DC blocker coefficient depend on sample rate
2024-04-09 08:00:35 -07:00
34be69df6d
Remove AudioDSPTools dependency. Do simple DC blocker instead of HPF.
2023-10-17 12:25:14 -07:00
8384827981
Removed test value.
2023-10-15 18:12:27 -07:00
e3c5b6bd4f
Update for normalization->loudness in NAM core.
2023-10-15 18:11:36 -07:00
eb46377d71
Update to latest NAM core. Handle model normalization with output gain.
2023-10-15 12:09:52 -07:00
b22f02c84e
Update NAM core. Switch to new simplified NAM process method.
2023-10-06 11:51:30 -07:00
46a73d83fa
Initialize model pre-run buffer to zero. Update to latest devel NAM Core with pre-warm instead of anti-pop.
2023-08-18 13:20:27 -07:00
6f1a423b5a
Optimize audio level loop code
...
Signed-off-by: falkTX <falktx@falktx.com >
2023-08-16 18:39:50 +02:00
fbf05b4472
Do 5Hz high pass filter to clean up any DC offset created by the model
2023-07-28 09:20:52 -07:00
d4482b3b14
Catch exceptions by reference
2023-07-12 09:50:27 -07:00
0255f36ae4
Don't report a worker error if we can't load a model since we're already logging an error.
2023-07-04 16:28:57 -07:00
207fb2281e
Fixed some warnings
2023-07-04 14:55:17 -07:00
dbf00f0ed3
Formatting
2023-07-04 14:53:39 -07:00
33a0c08327
Set currentModelPath in restore() - even if it is null or empty.
...
Set it to empty and clear the current model if a model fails to load.
2023-07-04 13:03:46 -07:00
739f43b151
state changed is no longer needed
...
Signed-off-by: falkTX <falktx@falktx.com >
2023-06-15 19:13:55 +02:00
eb35260789
Pre-run model to ensure all needed buffers are allocated in advance
...
Signed-off-by: falkTX <falktx@falktx.com >
2023-06-15 17:06:32 +02:00
4e9b7d6b49
rework worker implementation, dont rely on class variables
...
previous implementation was racy and bound to issues when more
than 1 file change request happened before worker was triggered.
using C++ move assignment is nice, but LV2 worker is a C API
that does not fit non-POD types very well, leading to awkward
implementations alike before with current + staged + deleted models.
let us "downgrade" to raw pointers, which are C compatible.
since LV2 worker rules are well defined, any crashes or racy
behaviour can be considered host-side bugs.
Signed-off-by: falkTX <falktx@falktx.com >
2023-06-15 17:00:02 +02:00
0d7954cf2a
General cleanup, no functional changes
...
Signed-off-by: falkTX <falktx@falktx.com >
2023-06-15 16:42:45 +02:00
3710dda839
Move fast tanh enable call to static plugin init
...
Signed-off-by: falkTX <falktx@falktx.com >
2023-06-15 16:36:31 +02:00
420cd4596d
Fix logger initialization to ensure correct operation if logging is not supported
2023-05-31 09:45:10 -07:00
41a45ace93
Float interface with NAM core. Workaround for Reaper crash. Use "nam" for model path fle type
2023-05-31 09:26:12 -07:00
207777ec3f
only do smoothing if we need to, and keep db conversion out of the sampe loop
2023-05-11 11:52:06 -07:00
d8193de9eb
do smoothing per-sample
2023-05-11 11:29:30 -07:00
10059d327c
parameter smoothing
2023-05-11 11:26:37 -07:00
3aaebfb25e
Fix notification port output.
2023-04-19 05:42:06 -04:00
6a1b7bfd2d
Additional checks for LV2 mapPath and freePath supprt
2023-04-11 08:26:40 -07:00
362b4b530d
Use lv2 freePath to free path data
2023-04-10 11:21:45 -07:00
6b21e67f88
Fixed incorrect model path size.
2023-04-10 10:41:28 -07:00
b12aec0876
Enable model loudness normalization. Update NeuralAmpModelerCore.
2023-04-10 10:26:01 -07:00
3a6abca1e6
lv2 error on failed model load
2023-04-10 10:17:19 -07:00
51f49b5fc1
Add back path free calls on linux only
2023-04-08 10:54:35 -07:00
022888bc6e
Merge branch 'main' of https://github.com/mikeoliphant/neural-amp-modeler-lv2
2023-04-03 15:10:48 -07:00
43f12ad73b
Re-enable fast tanh approximation
2023-04-03 15:10:33 -07:00
a7a90f1e6a
Remove path free() calls.
2023-04-03 10:03:02 -07:00
1c41097285
Added support to save/restore model file with presets/state
2023-04-02 15:24:32 -05:00
95ca0ad631
Add floating point denormal code
2023-03-24 11:00:51 -07:00
52942dc16b
Delete local NAM code
2023-03-23 12:09:33 -07:00
15642132ba
remove "rate" from constructor
2023-03-21 09:45:02 -07:00
cbf1ad7ccf
input and output level control ports
2023-03-13 13:16:10 -07:00
73975aeef6
Load models using LV2 work scheduler
2023-03-13 12:48:35 -07:00
99c709d70d
plugin cleanup
2023-03-13 09:52:15 -07:00
Mike Oliphant