Commit Graph

Select branches

Hide Pull Requests

dev

Mono Color

• 3d3fb14a4b Update README.md Eren Golge 2018-09-27 15:47:13 +02:00
• f48c3b7c1e Notebooks updated Eren 2018-09-27 15:35:06 +02:00
• 5496977abe Update README.md Eren Golge 2018-09-27 15:27:08 +02:00
• 3397ad39fa Add images and hard-sentences Eren 2018-09-27 15:13:04 +02:00
• df9746bc7c README update Eren 2018-09-27 15:10:15 +02:00
• 231607a546 Reduce batch group size Eren 2018-09-26 16:52:47 +02:00
• 006354320e Merge branch 'attention-smoothing' into attn-smoothing-bgs-sigmoid-wd Eren 2018-09-26 16:51:42 +02:00
• 3dda49f1c9 Merge branch 'larger-attn-larger-model-sigmoid' into attn-smoothing-bgs-sigmoid-wd Eren 2018-09-26 16:51:34 +02:00
• c89a3098dd Merge branch 'weight-decay' into attn-smoothing-bgs-sigmoid-wd Eren 2018-09-26 16:49:39 +02:00
• 95eb3367bd bug fix Eren 2018-09-21 21:51:38 +02:00
• 83b65557a9 bug fix Eren 2018-09-21 21:51:38 +02:00
• 607ca52019 bug fix Eren 2018-09-21 21:51:38 +02:00
• 41bfa95736 bug fix Eren 2018-09-21 21:51:38 +02:00
• c36a1f7deb Make audio folder and save audio with scipy Eren 2018-09-21 17:27:02 +02:00
• 172e0262de Make audio folder and save audio with scipy Eren 2018-09-21 17:27:02 +02:00
• 34eeaee58b Make audio folder and save audio with scipy Eren 2018-09-21 17:27:02 +02:00
• 8fe733c166 Make audio folder and save audio with scipy Eren 2018-09-21 17:27:02 +02:00
• 30fea0b957 Batch group shuffling Eren 2018-09-20 11:08:12 +02:00
• 6cfdf2d468 config change Eren 2018-09-19 15:22:43 +02:00
• f60e4497a6 apply sigmoid to outputs Eren 2018-09-15 18:05:34 +02:00
• e7278437ee Please enter the commit message for your changes. Lines starting Eren 2018-09-06 15:27:15 +02:00
• f2ef1ca36a Smmothed attention as in https://arxiv.org/pdf/1506.07503.pdf Eren 2018-09-19 15:08:43 +02:00
• d9deeacc7a Config Eren 2018-09-19 15:14:38 +02:00
• 16db5159f1 Weight decay described here: http://www.fast.ai/2018/07/02/adam-weight-decay/ Eren 2018-09-19 14:25:30 +02:00
• 9b29b4e281 Weight decay Eren 2018-09-18 12:56:07 +02:00
• a165cd7bda Bug fix audio saving Eren 2018-09-19 15:45:08 +02:00
• 10c2ff2074 config change Eren 2018-09-19 15:13:25 +02:00
• 67df385275 Explicit padding for unbalanced padding sizes Eren 2018-09-19 14:20:02 +02:00
• fd830c6416 Attention convolution padding correction for TF "SAME" Eren 2018-09-19 14:16:21 +02:00
• c52d3f16f9 Bug fix, prevent save_wav to modify given variable Eren 2018-09-19 14:05:10 +02:00
• 2bcd7dbb6f Change functional padding with padding layer Eren 2018-09-18 16:00:47 +02:00
• 00c0c9cde6 Padding with funcitonal interface to match TF "SAME" Eren 2018-09-17 20:19:09 +02:00
• 4a741c64b3 Loader text bug fix Eren 2018-09-17 18:58:48 +02:00
• 6a5414d4f6 Merge branch 'master' into larger-attn-larger-model Eren 2018-09-17 18:56:00 +02:00
• 41c06029b8 Merge branch 'larger-attn-larger-model' Eren 2018-09-17 13:58:57 +02:00
• 4b08c3a13f Save audio samples per validation Eren 2018-09-16 13:32:59 +02:00
• 48035a329f Remove redundant import Eren 2018-09-07 09:53:40 +02:00
• 28ef4a7acb README update Eren 2018-09-07 09:53:30 +02:00
• c78ec7dc36 remove redundant variables from tests Eren 2018-09-06 15:43:55 +02:00
• 443659a731 reverse post-cbhg size Eren 2018-09-06 15:31:07 +02:00
• f377cd3cb8 larger attention filter size and mode filters Eren 2018-09-06 15:27:15 +02:00
• 56c6d0cac8 Remove min max mel freq Eren 2018-09-06 15:25:47 +02:00
• 070156b631 config change Eren 2018-09-06 15:25:03 +02:00
• 7d66bdc5f4 update model size to paper Eren 2018-09-06 14:36:53 +02:00
• 8e564be3c7 config change Eren 2018-09-06 14:33:21 +02:00
• 03e1c06f5b bug fix Eren 2018-08-13 15:04:00 +02:00
• 73c3ab546d Testing update Eren 2018-08-13 15:02:30 +02:00
• a15b3ec9a1 pytorch 0.4.1update Eren 2018-08-13 15:02:17 +02:00
• 90b96f9bed config update Eren 2018-08-13 13:18:50 +02:00
• bb526c296f Change scheduler AnnealLR and catch audio synthesis error in eval time Eren 2018-08-13 13:13:45 +02:00
• f5e87a0c70 master update from loc-sens-attn-smaller-post-cbhg Eren 2018-08-12 15:07:56 +02:00
• 6818e11185 Make lr scheduler configurable Eren 2018-08-12 15:02:06 +02:00
• a518c02638 config update Eren 2018-08-12 15:01:56 +02:00
• f7add3c8e5 tensorboardx plotting figures Eren 2018-08-11 16:53:09 +02:00
• b65f5bd1d0 bug fix for plotting Eren 2018-08-11 16:09:47 +02:00
• 3b2654203d fixing size mismatch Eren 2018-08-10 18:47:09 +02:00
• 9100e5762a Disabling cappint gradient value Eren 2018-08-10 17:49:48 +02:00
• 9e2baa4f03 Dont tim model param names since we don't use DataPatallel initilization Eren 2018-08-10 17:49:21 +02:00
• 96e2e3c776 Use step wise LR scheduler + adapt train.py for passing squence mask directly Eren 2018-08-10 17:48:19 +02:00
• e0bce1d2d1 Pass mask instead of length to model Eren 2018-08-10 17:43:45 +02:00
• 2196ce9eba Config change Eren 2018-08-10 17:42:19 +02:00
• f1271b4ce8 Audio.py check and remove redundant code Eren G 2018-08-09 15:12:53 +02:00
• ecd31af125 Formating and printing more about the model Eren G 2018-08-08 18:45:02 +02:00
• d5febfb187 Setting up network size according to the reference paper Eren G 2018-08-08 12:34:44 +02:00
• 790a1b4639 config changes Eren G 2018-08-08 12:31:03 +02:00
• abf89b107b Update audio.py Eren G 2018-08-03 15:37:37 +02:00
• 59738e9676 Add versions to setup.py Eren G 2018-08-03 15:37:26 +02:00
• fc9ede5a28 Add versions to requirement file Eren G 2018-08-03 15:35:04 +02:00
• f5537dc48f pep8 format all Eren G 2018-08-02 16:34:17 +02:00
• 3238ffa3e6 audio_lws update, suppress warnings Eren G 2018-08-02 14:30:05 +02:00
• ed56dd68f2 LJSpeech config update Eren G 2018-08-02 14:29:38 +02:00
• 411d80efef kusal config change Eren G 2018-08-02 14:29:26 +02:00
• 908f6c86f3 Update config_kunsal Eren G 2018-07-30 14:10:01 +02:00
• bce43f4d6e Update text Eren G 2018-07-30 14:09:50 +02:00
• 3ad5aea93b Add Kusal training config Eren G 2018-07-30 14:04:08 +02:00
• dd512876cd path change wav kusal Eren G 2018-07-30 14:03:38 +02:00
• fa790db62e trim silence if necessary Eren G 2018-07-30 14:03:07 +02:00
• a39368b858 New dataset Kusal from Mycroft Eren G 2018-07-30 13:53:05 +02:00
• 8bc4fe8aac Printing fix with flush and spaceing Eren G 2018-07-30 13:52:39 +02:00
• 8864252941 Test updates for loaders Eren G 2018-07-30 13:52:16 +02:00
• 5d7338ddf6 Requirements change Eren G 2018-07-30 13:51:47 +02:00
• 62054da8f1 load audio change Eren G 2018-07-30 13:51:25 +02:00
• 560330e22e Small changex Eren G 2018-07-30 13:50:57 +02:00
• adedd7b1a9 Set num output units based on audio processor and set imported audio processor based on config.json Eren G 2018-07-27 16:14:10 +02:00
• 355df8fe9b Keep avg step time Eren G 2018-07-27 16:13:55 +02:00
• fcc21cdc69 Add lws to requrements Eren G 2018-07-27 16:12:58 +02:00
• 07291c0290 use tqdm with extact_feat Eren G 2018-07-27 16:12:45 +02:00
• faf6bdc4ca Make preemphasis configurable based on config.json Eren G 2018-07-27 16:12:25 +02:00
• b66f610bf4 Config audio processor based on config.json Eren G 2018-07-27 16:11:08 +02:00
• dbfa3e22a5 use tqdm extract_Feat Eren G 2018-07-27 16:10:52 +02:00
• a7f1be6b5c config.json update Eren G 2018-07-27 16:10:25 +02:00
• 06abb07b1e Update branch info Eren G 2018-07-27 16:08:53 +02:00
• 30561d0b5f Save best model based on train loss Eren G 2018-07-27 13:47:13 +02:00
• 13ace3a9e7 Remove tqdm from extract_feat Eren G 2018-07-27 13:48:01 +02:00
• 25b6769246 Some bug fixes Eren G 2018-07-26 13:33:05 +02:00
• e156738920 adding cache loader Eren G 2018-07-25 19:14:07 +02:00
• bb8aaadc1f Update branch info Eren G 2018-07-25 12:56:52 +02:00
• f24a9e74b8 Define metafiles in conf.json Eren G 2018-07-25 12:34:42 +02:00
• b83d125e52 Apply pre-emphasis again and disable min-max mel_freqs Eren G 2018-07-23 16:15:51 +02:00
• c72f309068 Plot specs and alignments for test sentences Eren G 2018-07-23 16:14:46 +02:00