Commit Graph

1734 Commits

Author SHA1 Message Date
Edresson f37159c135 add Jupyter Notebook for Extract Speaker Embedding per sample using AngleProto 2020-08-05 19:33:35 +02:00
Edresson bd4c6ee42a add speaker encoder parameters in config.json, for more easy changes 2020-08-05 19:33:35 +02:00
Edresson cb212b8049 update deprecated functions call from speaker encoder 2020-08-05 19:33:35 +02:00
Edresson 5e11d81e12 fix broken imports in speaker encoder 2020-08-05 19:33:08 +02:00
Edresson ac85ccae99 update speaker encoder config.json, for compatibility with the TTS model 2020-08-05 19:32:41 +02:00
Edresson bc09ca8111 add suport for AngleProto loss 2020-08-05 19:32:41 +02:00
Edresson f0bcc390d2 Implement Angular Prototypical loss 2020-08-05 19:28:44 +02:00
Edresson 8b9c951da7 Ops! Map Notebooks imports for TTS current version 2020-08-05 19:28:44 +02:00
Edresson ad570f3cab add Jupyter Notebook for Extract Speaker Embedding per sample using GE2E 2020-08-05 19:28:44 +02:00
Edresson 9d28f31168 add Colab Notebook from TTS Multi-Speaker with GST 2020-08-05 19:28:44 +02:00
Edresson 66452918b7 clean Colab Notebook outputs 2020-08-05 19:28:44 +02:00
Edresson d87c7227dd add Colab Notebook from TTS Multi-Speaker 2020-08-05 19:28:44 +02:00
Edresson ec2de5c88d added integrity test for GST dictionary in synthesis 2020-08-05 19:28:44 +02:00
Edresson 0fa2544fa5 bugfix in tacotron and tacotron 2 gst inference 2020-08-05 19:28:44 +02:00
Edresson d038a00602 add gst style dict suport in synthesize 2020-08-05 19:28:44 +02:00
Edresson 27ddef141d add gst suport in synthesize 2020-08-05 19:28:44 +02:00
Edresson ae62efd9eb Force the loading of the wav in synthesis using the AP sample rate to avoid breaking the demo 2020-08-05 19:28:44 +02:00
Edresson 496a4be3e3 add support for synthesize using variable size external embedding and add bugfix in scipy.io import 2020-08-05 19:28:44 +02:00
Edresson f91b9eeda1 fix Lint check 2020-08-05 19:26:42 +02:00
Edresson a2ee48c28e bugfix in GST 2020-08-05 19:26:42 +02:00
Edresson 89d338358e add External Embedding per sample instead of nn.Embedding 2020-08-05 19:26:42 +02:00
Edresson be77e24a39 bugfix in DDC now DDC work on Tacotron1 2020-08-05 19:23:18 +02:00
Edresson 70c665b9c4 add support fot VCTK and BRSpeech dataset 2020-08-05 19:19:42 +02:00
SanjaESC c4828b2b9e small gst config change 2020-08-05 19:19:07 +02:00
SanjaESC 1436206224 override compute_gst in tacotron2 model 2020-08-05 19:18:34 +02:00
SanjaESC 77bfb881d7 tacotrongst test + test fixes 2020-08-05 19:18:34 +02:00
SanjaESC 89918c6e53 pylint 2020-08-05 19:17:14 +02:00
SanjaESC 447176258c fix fft_size key error 2020-08-05 19:16:38 +02:00
SanjaESC c2d8a338a1 No need to query every token when none were passed 2020-08-05 19:16:19 +02:00
SanjaESC 84b7ab6ee6 Added support for Tacotron2 GST + abbility to condition style input with wav or tokens 2020-08-05 19:16:19 +02:00
erogol fe081d4f7c fixing rebase issues 2020-08-05 18:33:22 +02:00
erogol 5c752799ae linter update 2020-08-05 18:03:28 +02:00
Edresson 6e7f33c798 add support for synthesize using variable size external embedding and add bugfix in scipy.io import 2020-08-05 18:02:44 +02:00
Edresson def7e49f59 travis unit tests fix and add Tacotron and Tacotron 2 GST and MultiSpeaker Tests 2020-08-05 18:00:20 +02:00
Edresson 7c12e94ee4 fix Lint check 2020-08-05 17:57:57 +02:00
Edresson 1d73566e4e bugfix in GST 2020-08-05 17:57:57 +02:00
Edresson 8a1c113df6 add External Embedding per sample instead of nn.Embedding 2020-08-05 17:57:57 +02:00
Edresson e265810e8c bugfix in DDC now DDC work on Tacotron1 2020-08-05 17:51:31 +02:00
Edresson b750452782 add Portuguese Cleaner 2020-08-05 17:41:52 +02:00
Edresson 93a9cc4683 add support fot VCTK and BRSpeech dataset 2020-08-05 17:41:52 +02:00
Eren Gölge 571f5761c9 update comment 2020-08-05 17:41:52 +02:00
SanjaESC b0c4a77aea small gst config change 2020-08-05 17:41:52 +02:00
SanjaESC e206ff8a28 override compute_gst in tacotron2 model 2020-08-05 17:37:38 +02:00
SanjaESC eb51d5409a pylint 2020-08-05 17:36:36 +02:00
SanjaESC bdf6944665 fix fft_size key error 2020-08-05 17:36:36 +02:00
SanjaESC 8d0d4919fd No need to query every token when none were passed 2020-08-05 17:36:36 +02:00
erogol b96e74dd49 add multi-speaker arguments to the model def 2020-08-05 17:35:58 +02:00
erogol 8c07ae734c Merge branch 'dev' of https://github.com/mozilla/TTS into dev 2020-08-04 14:30:07 +02:00
erogol 0a34eb6f08 testing fix 2020-08-04 14:29:05 +02:00
erogol e386caa071 mass linter fix 2020-08-04 14:07:47 +02:00