104 Commits

Author	SHA1	Message	Date
Michael Hansen	bde9df4527	Merge branch 'master' of https://github.com/rhasspy/piper	2023-04-15 22:44:41 -05:00
Michael Hansen	cbed19e1b4	Using patched espeak-ng	2023-04-13 14:18:50 -05:00
Trung Le	48e4cbd6e9	Correcting librosa mel compute	2023-04-13 17:36:01 +07:00
Michael Hansen	5b64824eea	Bump version	2023-04-12 15:56:48 -05:00
Michael Hansen	e1d34f14fb	Split into sentences and output audio as available	2023-04-12 15:56:06 -05:00
Michael Hansen	f8386b1984	Use nanoseconds for timestamp	2023-04-11 12:06:39 -05:00
Heewon Lee	476e79302d	Change the mycroft dataset format	2023-04-11 21:12:10 +09:00
Michael Hansen	db93d62080	Onnxruntime 1.14	2023-04-10 17:14:56 -05:00
Michael Hansen	28a45a2a46	Fix RTF	2023-04-10 15:42:55 -05:00
Michael Hansen	c53f739f5f	Add logging	2023-04-10 15:37:56 -05:00
Michael Hansen	92eb23197b	Add stats	2023-04-10 15:23:48 -05:00
Michael Hansen	c945c141f1	More benchmarks	2023-04-10 14:37:34 -05:00
Michael Hansen	49dd00cc45	Merge branch 'master' of pumpkin.local:Mike/Projects/larynx2	2023-04-10 13:16:37 -05:00
Michael Hansen	1eed98ecd9	Add benchmark	2023-04-09 20:15:05 -05:00
Michael Hansen	3d5d53f07c	Remove executor	2023-04-09 11:38:11 -05:00
Michael Hansen	d88965181b	Add timeout to phonemize	2023-04-05 17:24:52 -05:00
Michael Hansen	70afec58bc	Rename to piper	2023-03-26 21:42:04 -05:00
Michael Hansen	3dfa161ba5	Add voice conversion script	2023-03-17 12:03:50 -05:00
Michael Hansen	44c2cf1ee9	Add --speaker-id to preprocess to force a speaker id	2023-03-17 12:03:37 -05:00
Michael Hansen	8873638771	Add --quality	2023-03-11 21:37:58 -06:00
Michael Hansen	894e5a949d	Use opset 15 for older PyTorch	2023-03-06 09:36:11 -06:00
Michael Hansen	48c7cc99f9	Add remaining args	2023-02-27 10:27:15 -06:00
Michael Hansen	b21d815298	Starting on Python inference	2023-02-26 23:03:11 -06:00
Michael Hansen	93d3744614	Include discriminator loss in validation	2023-02-25 20:08:50 -06:00
Michael Hansen	b8e3058d7a	Add --checkpoint-epochs	2023-02-24 15:03:02 -06:00
Michael Hansen	657a1fae74	Add --single-speaker to preprocess	2023-02-24 12:54:47 -06:00
Michael Hansen	5493e5dd4c	Fix up docker	2023-02-24 12:45:57 -06:00
Michael Hansen	131dbf4b5b	Don't load dataset during inference	2023-02-23 09:22:55 -06:00
Michael Hansen	aab406cfdd	Small clean up	2023-02-14 10:13:18 -06:00
Michael Hansen	acc3068176	Export without loading dataset	2023-02-14 09:03:06 -06:00
Michael Hansen	b2af638072	Upgrade to onnxruntime 1.13.1	2023-01-14 12:09:23 -06:00
Michael Hansen	afbcfb4df3	Fix WAV length	2023-01-11 17:42:23 -06:00
Michael Hansen	1d858e8128	Use RPATH in docker build	2023-01-10 16:03:50 -06:00
Michael Hansen	4c410fed2c	Add scale parameters	2023-01-10 14:11:38 -06:00
Michael Hansen	a34b2d4f3b	Fix synthesis speaker id bug	2023-01-10 14:01:47 -06:00
Michael Hansen	dfa03d80e6	Before consolidated main	2023-01-09 16:35:32 -06:00
Michael Hansen	6c949aa5c5	Add generator export/infer	2023-01-08 21:57:53 -06:00
Michael Hansen	e5062c9496	Exporting to TorchScript	2023-01-08 21:38:54 -06:00
Michael Hansen	f1cc4e58bd	Unsupported value kind: Tensor	2023-01-07 11:20:19 -06:00
Michael Hansen	bebc36014a	Add torchscript export	2023-01-07 10:57:34 -06:00
Michael Hansen	a6d72d7990	Add speaker id to infer_onnx	2023-01-05 23:14:08 -05:00
Michael Hansen	a7fe73390e	Add multispeaker	2023-01-05 21:47:08 -06:00
Michael Hansen	06a154a4ed	Pass speaker id during verification	2023-01-05 17:23:02 -05:00
Michael Hansen	c38020cb14	Fix speaker ids	2023-01-03 00:13:52 -05:00
Michael Hansen	56b09d4019	Assign speaker ids by number of utterances	2023-01-02 22:04:23 -05:00
Michael Hansen	3c518d65b7	Try relative file path for WAV in ljspeech	2023-01-02 21:51:24 -05:00
Michael Hansen	5f704228f6	Add --max-phoneme-ids	2022-11-11 17:55:04 -05:00
Michael Hansen	11b294a461	Clean up export onnx	2022-11-11 16:59:57 -05:00
Michael Hansen	042718a481	Add bos/eos and pad intersperse	2022-11-11 15:59:34 -06:00
Michael Hansen	7c22049330	Use multiprocess in preprocess script	2022-11-11 16:42:48 -05:00