rafaelvalle
|
825ffa47d1
|
inference.ipynb: reverting fp16 inference for now
|
6 years ago |
rafaelvalle
|
4d7b04120a
|
inference.ipynb: changing waverglow inference fo fp16
|
6 years ago |
rafaelvalle
|
6e430556bd
|
train.py: val logger on gpu 0 only
|
6 years ago |
rafaelvalle
|
3973b3e495
|
hparams.py: distributed using tcp
|
6 years ago |
rafaelvalle
|
52a30bb7b6
|
distributed.py: replacing to avoid distributed error
|
6 years ago |
rafaelvalle
|
0ad65cc053
|
train.py: renaming variable to n_gpus
|
6 years ago |
rafaelvalle
|
8300844fa7
|
hparams.py: removing 22khz
|
6 years ago |
rafaelvalle
|
f06063f746
|
train.py: renaming function, removing dataparallel
|
6 years ago |
rafaelvalle
|
3045ba125b
|
inference.ipynb: cleanup
|
6 years ago |
rafaelvalle
|
4c4aca3662
|
README.md: layout
|
6 years ago |
rafaelvalle
|
05dd8f91d2
|
README.md: adding submodule init to README
|
6 years ago |
rafaelvalle
|
5d66c3deab
|
adding waveglow submodule
|
6 years ago |
Rafael Valle
|
f02704f338
|
Merge pull request #96 from NVIDIA/clean_slate
Clean slate
|
6 years ago |
rafaelvalle
|
ba8cf36198
|
requirements.txt: removing pytorch 0.4 from requirements. upgrading to 1.0
|
6 years ago |
rafaelvalle
|
b5e0a93946
|
inference.ipynb: updating inference file with relative paths
|
6 years ago |
rafaelvalle
|
58b0ec61bd
|
README.md: updating requirements and inference demo
|
6 years ago |
rafaelvalle
|
1ad939df1a
|
inference.ipynb: setting relative model paths
|
6 years ago |
rafaelvalle
|
32b9a135d0
|
utils.py: updating
|
6 years ago |
rafaelvalle
|
ce29e13959
|
train.py: updating
|
6 years ago |
rafaelvalle
|
1ea6ed5861
|
text/symbols.py: updating symbols
|
6 years ago |
rafaelvalle
|
cdfde985e5
|
text/__init__.py: remove stop token
|
6 years ago |
rafaelvalle
|
e314bb4cd0
|
stft.py: fix filter winlength error
|
6 years ago |
rafaelvalle
|
4af4ccb135
|
model.py: rewrite
|
6 years ago |
rafaelvalle
|
1ec0e5e8cd
|
layers.py: rewrite
|
6 years ago |
rafaelvalle
|
249afd8043
|
inference.ipynb: import taco2model to be public
|
6 years ago |
rafaelvalle
|
1b243d5d5a
|
hparams.py:rewrite
|
6 years ago |
rafaelvalle
|
d0aa9e7d32
|
distributed.py: rewrite
|
6 years ago |
rafaelvalle
|
1683a57ae5
|
data_utils.py: rewrite
|
6 years ago |
Rafael Valle
|
fc0cf6a89a
|
Merge pull request #53 from cobr123/patch-1
add pillow
|
6 years ago |
cobr123
|
8de38495be
|
add pillow
|
6 years ago |
rafaelvalle
|
7eb045206c
|
README.md: updating readme to explicitly mention that mel representation of WaveNet and Tacotron2 must be the same
|
6 years ago |
rafaelvalle
|
c67005f1be
|
Dockerfile: adding jupyter to dockerfile
|
6 years ago |
rafaelvalle
|
cb3794796f
|
Dockerfile: removing return from Dockerfile
|
6 years ago |
Rafael Valle
|
a8de973923
|
Merge pull request #37 from yoks/master
`used_saved_learning_rate` fix
|
6 years ago |
yoks
|
a0ae2da05f
|
`used_saved_learning_rate` fix
`used_saved_learning_rate` name change to `use_saved_learning_rate`
|
6 years ago |
rafaelvalle
|
34066ac4fc
|
requirements.txt: setting torch to 0.4.0
|
6 years ago |
rafaelvalle
|
12ab5ba89c
|
model.py: setting weight initialization to xavier uniform
|
6 years ago |
rafaelvalle
|
d10da5f41e
|
hparams.py: commenting n_frames_per_step to indicate that currently only 1 frame per step is supported now
|
6 years ago |
rafaelvalle
|
5f0ea06c41
|
hparams.py: adding use saved learning rate param
|
6 years ago |
rafaelvalle
|
22bcff1155
|
hparams.py: adding use saved learning rate param
|
6 years ago |
rafaelvalle
|
2e934478a4
|
README.md: being explicit about action
|
6 years ago |
rafaelvalle
|
8ae231b10b
|
README.md: more explicit about demo audio
|
6 years ago |
rafaelvalle
|
4d733d1bdd
|
README.md: including demo.wav in readme
|
6 years ago |
rafaelvalle
|
b4e52404ab
|
adding demo.wav file
|
6 years ago |
Rafael Valle
|
064629c9bc
|
Merge pull request #23 from NVIDIA/attention_full_mel
model.py: attending to full mel instead of prenet and dropout mel
|
6 years ago |
Rafael Valle
|
d5b64729d1
|
model.py: moving for better readibility
|
6 years ago |
Rafael Valle
|
977cb37cea
|
model.py: attending to full mel instead of prenet and dropout mel
|
6 years ago |
Rafael Valle
|
da30fd8709
|
Merge pull request #20 from NVIDIA/fp16_path
Fp16 patch, not path!
|
6 years ago |
Rafael Valle
|
27b1767cb2
|
train.py: fixing typo
|
6 years ago |
Rafael Valle
|
817cd403d4
|
Merge branch 'master' of https://github.com/NVIDIA/tacotron2 into load_mel_from_disk
|
6 years ago |