Commit Graph

571 Commits

Author SHA1 Message Date
chenfeiyu 3e8a156348 add synthesis script for pwg 2021-06-26 19:31:52 +08:00
chenfeiyu 577c3b4f10 add use_mfa example 2021-06-25 07:31:40 +00:00
chenfeiyu 7e049a7744 add expanuser for preprocess, updater backward testing for pwg 2021-06-25 01:01:30 +08:00
chenfeiyu 542bbf6a81 add profiling tool and a config for parallel training 2021-06-24 14:53:38 +00:00
liangyunming ea5cb8e71f Dynamic to static 2021-06-23 15:18:37 +08:00
chenfeiyu b5f99a925f add numerical testing for PWGGenerator 2021-06-21 11:17:01 +00:00
chenfeiyu 83c9f0aeae add snapshot and visualizer 2021-06-21 09:56:26 +00:00
chenfeiyu a738954001 1. change default data layout to channel last in preprocessing;
2. add Summary and DictSummary for aggrelation of evaluation losses;
3. add unittest for report ans scope.
2021-06-18 09:44:32 +00:00
chenfeiyu 3977632b07 add docstrings to pwg 2021-06-18 03:21:02 +00:00
chenfeiyu 7ac0d3ce12 add docstring to pwg 2021-06-18 03:09:46 +00:00
chenfeiyu fbc7e51fc9 1. add compute_statistics and normalize;
2. use jsonlines to read and write metadata by default;
3. use threadpool to replace processpool in preprocessing cause it is faster.
2021-06-18 02:49:49 +00:00
chenfeiyu 30045cf602 move synchronize up before timer starts 2021-06-17 03:12:35 +00:00
chenfeiyu 58a988c789 add code to compute statistics 2021-06-16 16:18:13 +00:00
chenfeiyu 27d3585606 add some profiling to unittesting 2021-06-16 14:42:11 +00:00
chenfeiyu 683cc1d30f add unit testings back 2021-06-16 13:49:18 +00:00
chenfeiyu 042e02d242 use paddle's profiler 2021-06-16 13:43:13 +00:00
chenfeiyu 8dbcc9bccb add profiling 2021-06-16 09:40:47 +00:00
Hui Zhang d96e2828b8
Merge pull request #119 from iclementine/fastspeech
add models/fastspeech2
2021-06-16 14:55:50 +08:00
chenfeiyu bbbe5a8b50 add profiling 2021-06-16 14:47:30 +08:00
chenfeiyu 643dfa0fa0 WIP: add models/fastspeech2 2021-06-16 10:57:53 +08:00
chenfeiyu 95f64c4f02 WIP: add some trainig info 2021-06-14 17:21:45 +08:00
chenfeiyu b0983e4d76 WIP: pwg training works 2021-06-14 17:05:37 +08:00
chenfeiyu 54c7905f40 WIP: training setup done 2021-06-13 17:24:15 +08:00
chenfeiyu 0067851950 fix imports 2021-06-11 05:26:22 +08:00
chenfeiyu 66062d29e5 WIP: add sample code for parallel wavegan 2021-06-11 02:43:10 +08:00
chenfeiyu 258083aea9 fix file names 2021-06-11 01:01:43 +08:00
chenfeiyu f0a5ac8c5a move docstring to class 2021-06-11 00:59:07 +08:00
chenfeiyu 13ab0bd608 remove task-specific Datasets and add a general purpose DataTable 2021-06-10 22:58:53 +08:00
chenfeiyu 3bf2e71734 checkpoint utility: add support for str in addition to Path object 2021-06-10 20:34:59 +08:00
chenfeiyu 60c16dcfb7 add AudioMelDataset for training vocoders. 2021-06-10 20:16:14 +08:00
chenfeiyu 988d6d3268 add h5py utility to interact with numpy 2021-06-10 18:57:46 +08:00
chenfeiyu 8e31783b51 Merge branch 'develop' of https://github.com/PaddlePaddle/Parakeet into pwg 2021-06-10 16:26:34 +08:00
chenfeiyu 6a8b3f92df add AudioDataset and MelDataset 2021-06-10 16:26:09 +08:00
Feiyu Chan a0763986e2
add kbest (#114)
* add kbest

* fix typos

* remove unnecessary imports

* fix del fn

* fix typos

* add k-latest
2021-06-10 10:36:14 +08:00
chenfeiyu 3c964fde54 add parallel wavegan model 2021-06-10 04:08:05 +08:00
chenfeiyu 759999c738 STFT and MelScale: register filters as buffer. 2021-06-10 04:06:06 +08:00
chenfeiyu c306f5c2b3 add k-latest 2021-06-10 03:39:54 +08:00
chenfeiyu dc9040dd4d fix typos 2021-06-10 02:45:45 +08:00
chenfeiyu 0114a808a2 fix del fn 2021-06-10 00:40:54 +08:00
chenfeiyu 13323bdf6a remove unnecessary imports 2021-06-10 00:32:49 +08:00
chenfeiyu 37a66f1506 fix typos 2021-06-10 00:29:25 +08:00
chenfeiyu b571b506c3 add kbest 2021-06-10 00:25:00 +08:00
Feiyu Chan 87a0cec0c7
Merge pull request #111 from iclementine/develop
fix a config key error
2021-05-18 18:59:53 +08:00
chenfeiyu 3aa6aed0fb fix a config key error 2021-05-18 18:57:13 +08:00
Feiyu Chan 4f288a6d4f
add ge2e and tacotron2_aishell3 example (#107)
* hacky thing, add tone support for acoustic model

* fix experiments for waveflow and wavenet, only write visual log in rank-0

* use emb add in tacotron2

* 1. remove space from numericalized representation;
2. fix decoder paddign mask's unsqueeze dim.

* remove bn in postnet

* refactoring code

* add an option to normalize volume when loading audio.

* add an embedding layer.

* 1. change the default min value of LogMagnitude to 1e-5;
2. remove stop logit prediction from tacotron2 model.

* WIP: baker

* add ge2e

* fix lstm speaker encoder

* fix lstm speaker encoder

* fix speaker encoder and add support for 2 more datasets

* simplify visualization code

* add a simple strategy to support multispeaker for tacotron.

* add vctk example for refactored tacotron

* fix indentation

* fix class name

* fix visualizer

* fix root path

* fix root path

* fix root path

* fix typos

* fix bugs

* fix text log extention name

* add example for baker and aishell3

* update experiment and display

* format code for tacotron_vctk, add plot_waveform to display

* add new trainer

* minor fix

* add global condition support for tacotron2

* add gst layer

* add 2 frontend

* fix fmax for example/waveflow

* update collate function, data loader not does not convert nested list into numpy array.

* WIP: add hifigan

* WIP:update hifigan

* change stft to use conv1d

* add audio datasets

* change batch_text_id, batch_spec, batch_wav to include valid lengths in the returned value

* change wavenet to use on-the-fly prepeocessing

* fix typos

* resolve conflict

* remove imports that are removed

* remove files not included in this release

* remove imports to deleted modules

* move tacotron2_msp

* clean code

* fix argument order

* fix argument name

* clean code for data processing

* WIP: add README

* add more details to thr README, fix some preprocess scripts

* add voice cloning notebook

* add an optional to alter the loss and model structure of tacotron2, add an alternative config

* add plot_multiple_attentions and update visualization code in transformer_tts

* format code

* remove tacotron2_msp

* update tacotron2 from_pretrained, update setup.py

* update tacotron2

* update tacotron_aishell3's README

* add images for exampels/tacotron2_aishell3's README

* update README for examples/ge2e

* add STFT back

* add extra_config keys into the default config of tacotron

* fix typos and docs

* update README and doc

* update docstrings for tacotron

* update doc

* update README

* add links to downlaod pretrained models

* refine READMEs and clean code

* add praatio into requirements for running the experiments

* format code with pre-commit

* simplify text processing code and update notebook
2021-05-13 17:49:50 +08:00
Hui Zhang 0aa7088d36
Merge pull request #97 from iclementine/develop
fix multiprocess training
2021-02-23 10:52:01 +08:00
chenfeiyu 6f1c534557 use exist_ok to ensure no conflict 2021-02-23 10:50:57 +08:00
chenfeiyu c6504ade5a Merge branch 'develop' of https://github.com/PaddlePaddle/Parakeet into develop 2021-02-23 10:40:26 +08:00
chenfeiyu e69ab88fe6 fix multiprocessing training: other processes have to wait untils the output directory in created. 2021-02-23 10:40:14 +08:00
Feiyu Chan dc055bde0a
Merge pull request #96 from iclementine/develop
fix experiments for waveflow and wavenet
2021-02-22 14:22:42 +08:00