114 commits

Author	SHA1	Message	Date
Jean-Marc Valin	df02d25f0c	Updated FARGAN model with softquant	2025-03-29 11:12:24 -04:00
Jean-Marc Valin	af6dbd84a8	Adds --chunks-per-offset option to train_rdovae.py	2025-03-28 20:04:51 -04:00
Jan Buethe	1ca6933ac4	added soft quantization to RDOVAE and FARGAN	2025-03-28 20:04:51 -04:00
Jean-Marc Valin	ebccedd918	Fix rdovae exporting	2025-03-28 20:04:50 -04:00
Jean-Marc Valin	b5aad6a282	Better model building instructions	2025-03-18 14:01:52 -04:00
Jan Buethe	7db26934e4	fix for buffer size calculation in osce.c ... update to osce evaluation script	2024-10-16 11:01:21 +02:00
Jan Buethe	fc96bb9cc7	added osce testing related scripts (ietf120)	2024-07-26 12:32:53 +02:00
Jean-Marc Valin	0e564fdfaf	More fixes for the non-blob weight export	2024-05-18 02:40:53 -04:00
Jan Buethe	1711e97165	fixed enable_binary_blob option for CWriter	2024-05-06 14:11:59 +02:00
Jan Buethe	20568812ae	README related osce fixes III	2024-04-29 17:13:45 +02:00
Jan Buethe	f6f2a328a6	README related osce fixes II	2024-04-29 15:58:03 +02:00
Jan Buethe	fe9ded3bd7	README related osce fixes	2024-04-29 15:56:46 +02:00
Jan Buethe	0dc559f060	added some bwe-related stuff	2024-04-24 12:17:51 +02:00
Jean-Marc Valin	ddbe483839	Update DRED REDME.md	2024-04-10 16:38:50 -04:00
Jean-Marc Valin	04347e67c6	Missing blockquotes	2024-04-10 02:02:59 -04:00
Jean-Marc Valin	acb7e0012e	FARGAN instructions	2024-04-10 02:01:40 -04:00
dependabot[bot]	d7fab5e892	Bump gitpython from 3.1.36 to 3.1.41 in /dnn/torch/osce ... Bumps [gitpython](https://github.com/gitpython-developers/GitPython) from 3.1.36 to 3.1.41. - [Release notes](https://github.com/gitpython-developers/GitPython/releases) - [Changelog](https://github.com/gitpython-developers/GitPython/blob/main/CHANGES) - [Commits](https://github.com/gitpython-developers/GitPython/compare/3.1.36...3.1.41) Signed-off-by: Jean-Marc Valin <jmvalin@jmvalin.ca>	2024-03-12 13:34:10 -04:00
Jan Buethe	cb1956d1fe	added list of training/validation files used for osce training	2024-03-07 15:15:37 +01:00
Jan Buethe	367a487e7c	hangover fix in osce/utils/pitch.py	2024-02-15 15:39:30 +01:00
Jan Buethe	735117b6d7	disabled sparse option in osce export script	2024-02-15 15:25:06 +01:00
Jan Buethe	7df2c67be1	fixes in osce python code	2024-01-23 17:10:34 +01:00
Jan Buethe	ec04a94eb2	bugfix in SilkFeatureNetPL	2024-01-22 15:12:52 +01:00
Jean-Marc Valin	1ddfcfd48c	Using PyTorch model (same architecture for now)	2024-01-21 02:11:50 -05:00
Jan Buethe	299e38cab7	Updated LACE and NoLACE models to version 2	2024-01-20 14:44:22 +01:00
Jean-Marc Valin	4f311a1ad4	PLC export script ... mostly untested	2024-01-17 02:26:48 -05:00
Jean-Marc Valin	26ddfd7135	PyTorch code for training the PLC model ... Should match the TF2 code, but mostly untested	2024-01-15 18:11:47 -05:00
Jean-Marc Valin	bd2e9a34fb	Add simulated loss to opus_demo	2023-12-22 03:42:16 -05:00
Jean-Marc Valin	bd710e97f3	C code for packet loss simulator	2023-12-21 21:30:53 -05:00
Jean-Marc Valin	b923fd1e28	lossgen: better training, README.md	2023-12-21 18:01:57 -05:00
Jean-Marc Valin	c40add59af	lossgen: can now dump weights	2023-12-21 16:57:35 -05:00
Jean-Marc Valin	627aa7f5b3	Packet loss generation model	2023-12-21 15:34:33 -05:00
Jan Buethe	7d328f5bfa	Merge LACE/NoLACE under OSCE framework	2023-12-20 03:42:44 -05:00
Jean-Marc Valin	7f7b2a1c66	Smaller version of fargan ... 800k parameters, 600 MFLOPS, with a receptive field of 3 feature vectors	2023-11-16 02:06:14 -05:00
Jean-Marc Valin	b0620c0bf9	Using sparse GRUs in DRED decoder ... Saves ~270 kB of weights in the decoder	2023-11-15 04:08:50 -05:00
Jean-Marc Valin	77594bf158	Dumping RDOVAE stats from XML	2023-11-08 17:32:43 -05:00
Jean-Marc Valin	222662dac8	DRED: quantize scale and dead zone to 8 bits	2023-11-07 18:10:50 -05:00
Jan Buethe	4e104555e9	added weight export script for LACE/NoLACE	2023-11-07 15:12:12 +01:00
Jan Buethe	8af5c6b4a1	added transposed 1d convolutions to wexchange	2023-11-07 11:54:22 +01:00
Jean-Marc Valin	0ab0640d4a	Split stats in two and remove useless dimensions	2023-11-07 00:07:14 -05:00
Jan Buethe	2386a60ec6	updated moc to match results in ietf118 presentation	2023-11-06 17:50:48 +01:00
Jean-Marc Valin	544b3e576c	DRED: quantize r and p0 parameters with 8 bits ... Only code non-degenerate symbols, which makes the encoder faster	2023-11-06 03:16:43 -05:00
Jan Buethe	da60266f6e	updated moc method	2023-11-02 16:52:50 +01:00
Jean-Marc Valin	ddd5669e79	Pitch and fargan model updates ... Removing one of the 2d conv layers for pitch estimation reduces complexity without noticeable degradation. FARGAN model has more adversarial training. Also, no need for the double precision in the low-pass filter.	2023-10-28 23:33:47 -04:00
Jean-Marc Valin	4598fe5409	Quantizing pitchdnn and rdovae weights	2023-10-20 12:54:13 -04:00
Jan Buethe	290be25b98	added 16kHz version of opus_compare in python	2023-10-20 14:24:27 +02:00
Jan Buethe	1accd2472e	finalized quantization option in export_rdovae_weights.py	2023-10-20 14:14:31 +02:00
Jean-Marc Valin	d720955d61	Marking RDOVAE layers to quantize	2023-10-19 16:06:52 -04:00
Jan Buethe	60ac1c6c99	prepared quantization implementation for DRED	2023-10-19 21:54:39 +02:00
Jan Buethe	2192e85b91	restructured osce readme	2023-10-19 21:45:45 +02:00
Jan Buethe	055c683018	added LACE/NoLACE checkpoint URL	2023-10-19 21:34:13 +02:00