RayeRen commited on
Commit
e931f78
1 Parent(s): 53fa903
docs/fastspeech2.md DELETED
@@ -1,53 +0,0 @@
1
- # Run FastSpeech 2
2
-
3
- ## Quick Start
4
-
5
- ### Install Dependencies
6
-
7
- Install dependencies following [readme.md](../readme.md)
8
-
9
- ### Set Config Path and Experiment Name
10
-
11
- ```bash
12
- export CONFIG_NAME=egs/datasets/audio/lj/fs2_orig.yaml
13
- export MY_EXP_NAME=fs2_exp
14
- ```
15
-
16
- ### Preprocess and binary dataset
17
-
18
- Prepare dataset following [prepare_data.md](./prepare_data.md)
19
-
20
- ### Prepare Vocoder
21
-
22
- Prepare vocoder following [prepare_vocoder.md](./prepare_vocoder.md)
23
-
24
- ## Training
25
-
26
- ```bash
27
- CUDA_VISIBLE_DEVICES=0 python tasks/run.py --config $CONFIG_NAME --exp_name $MY_EXP_NAME --reset
28
- ```
29
-
30
- You can check the training and validation curves open Tensorboard via:
31
-
32
- ```bash
33
- tensorboard --logdir checkpoints/$MY_EXP_NAME
34
- ```
35
-
36
- ## Inference (Testing)
37
-
38
- ```bash
39
- CUDA_VISIBLE_DEVICES=0 python tasks/run.py --config $CONFIG_NAME --exp_name $MY_EXP_NAME --infer
40
- ```
41
-
42
- ## Citation
43
-
44
- If you find this useful for your research, please use the following.
45
-
46
- ```
47
- @inproceedings{ren2020fastspeech,
48
- title={FastSpeech 2: Fast and High-Quality End-to-End Text to Speech},
49
- author={Ren, Yi and Hu, Chenxu and Tan, Xu and Qin, Tao and Zhao, Sheng and Zhao, Zhou and Liu, Tie-Yan},
50
- booktitle={International Conference on Learning Representations},
51
- year={2020}
52
- }
53
- ```
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
inference/tts/gradio/gradio_settings.yaml CHANGED
@@ -1,12 +1,12 @@
1
- title: 'NATSpeech/PortaSpeech'
2
  description: |
3
- Gradio demo for NATSpeech/PortaSpeech. To use it, simply add your audio, or click one of the examples to load them. Note: This space is running on CPU, inference times will be higher.
4
  article: |
5
- Link to <a href='https://github.com/NATSpeech/NATSpeech/blob/main/docs/portaspeech.md' style='color:blue;' target='_blank\'>Github REPO</a>
6
  example_inputs:
7
  - |-
8
  the invention of movable metal letters in the middle of the fifteenth century may justly be considered as the invention of the art of printing.
9
  - |-
10
  produced the block books, which were the immediate predecessors of the true printed book,
11
- inference_cls: inference.tts.ps_flow.PortaSpeechFlowInfer
12
- exp_name: ps_normal_exp
 
1
+ title: 'NATSpeech/DiffSpeech'
2
  description: |
3
+ Gradio demo for NATSpeech/DiffSpeech. To use it, simply add your audio, or click one of the examples to load them. Note: This space is running on CPU, inference times will be higher.
4
  article: |
5
+ Link to <a href='https://github.com/NATSpeech/NATSpeech/blob/main/docs/diffspeech.md' style='color:blue;' target='_blank\'>Github REPO</a>
6
  example_inputs:
7
  - |-
8
  the invention of movable metal letters in the middle of the fifteenth century may justly be considered as the invention of the art of printing.
9
  - |-
10
  produced the block books, which were the immediate predecessors of the true printed book,
11
+ inference_cls: inference.tts.ds.DiffSpeechInfer
12
+ exp_name: diffsinger