Skip to content

D
DeepSpeech

PaddlePaddle / DeepSpeech
大约 2 年 前同步成功

通知 210
Star 8425
Fork 1598
D
DeepSpeech
未验证 提交 0fa3fdb9 编写于 作者: u010070587's avatar u010070587 提交者： GitHub
浏览文件
操作

Merge pull request #2068 from yt605155624/p_norm 

[TTS]add onnx models for aishell3/ljspeech/vctk's tts3/voc1/voc5
上级 ec759094 6a45c5c3
显示空白变更内容
内联 并排
Showing with 499 addition and 86 deletion
README_cn.md
浏览文件 @ 0fa3fdb9
......@@ -691,7 +691,7 @@ PaddleSpeech 的 **语音合成** 主要包含三个模块：文本前端、声
- 非常感谢 [phecda-xu](https://github.com/phecda-xu)/[PaddleDubbing](https://github.com/phecda-xu/PaddleDubbing) 基于 PaddleSpeech 的 TTS 模型搭建带 GUI 操作界面的配音工具。
- 非常感谢 [jerryuhoo](https://github.com/jerryuhoo)/[VTuberTalk](https://github.com/jerryuhoo/VTuberTalk) 基于 PaddleSpeech 的 TTS GUI 界面和基于 ASR 制作数据集的相关代码。
- 非常感谢 [vpegasus](https://github.com/vpegasus)/[xuesebot](https://github.com/vpegasus/xuesebot) 基于 PaddleSpeech 的 ASR与TTS 设计的可听、说对话机器人
- 非常感谢 [vpegasus](https://github.com/vpegasus)/[xuesebot](https://github.com/vpegasus/xuesebot) 基于 PaddleSpeech 的 ASR 与 TTS 设计的可听、说对话机器人。
此外，PaddleSpeech 依赖于许多开源存储库。有关更多信息，请参阅 [references](./docs/source/reference.md)。
......
docs/source/install_cn.md
浏览文件 @ 0fa3fdb9
......@@ -116,7 +116,7 @@ conda install -y -c gcc_linux-64=8.4.0 gxx_linux-64=8.4.0
python3 -m pip install paddlepaddle-gpu==2.2.0 -i https://mirror.baidu.com/pypi/simple
```
### 安装 PaddleSpeech
最后安装 `paddlespeech`，这样你就可以使用 `paddlespeech`中已有的 examples：
最后安装 `paddlespeech`，这样你就可以使用 `paddlespeech` 中已有的 examples：
```bash
# 部分用户系统由于默认源的问题，安装中会出现kaldiio安转出错的问题，建议首先安装pytest-runner:
pip install pytest-runner -i https://pypi.tuna.tsinghua.edu.cn/simple
......@@ -137,7 +137,7 @@ Docker 是一种开源工具，用于在和系统本身环境相隔离的环境
在 [Docker Hub](https://hub.docker.com/repository/docker/paddlecloud/paddlespeech) 中获取这些镜像及相应的使用指南，包括 CPU、GPU、ROCm 版本。
如果您对自动化制作docker镜像感兴趣，或有自定义需求，请访问 [PaddlePaddle/PaddleCloud](https://github.com/PaddlePaddle/PaddleCloud/tree/main/tekton) 做进一步了解。
如果您对自动化制作 docker 镜像感兴趣，或有自定义需求，请访问 [PaddlePaddle/PaddleCloud](https://github.com/PaddlePaddle/PaddleCloud/tree/main/tekton) 做进一步了解。
完成这些以后，你就可以在 docker 容器中执行训练、推理和超参 fine-tune。
### 选择2： 使用有 root 权限的 Ubuntu
- 使用apt安装 `build-essential`
......@@ -173,7 +173,7 @@ conda install -y -c conda-forge sox libsndfile swig bzip2 libflac bc
python3 -m pip install paddlepaddle-gpu==2.2.0 -i https://mirror.baidu.com/pypi/simple
```
### 用开发者模式安装 PaddleSpeech
部分用户系统由于默认源的问题，安装中会出现kaldiio安转出错的问题，建议首先安装pytest-runner:
部分用户系统由于默认源的问题，安装中会出现 kaldiio 安转出错的问题，建议首先安装 pytest-runner:
```bash
pip install pytest-runner -i https://pypi.tuna.tsinghua.edu.cn/simple
```
......
docs/source/released_model.md
浏览文件 @ 0fa3fdb9
# Released Models
## Speech-to-Text Models
......@@ -34,32 +33,33 @@ Language Model | Training Data | Token-based | Size | Descriptions
## Text-to-Speech Models
### Acoustic Models
Model Type | Dataset| Example Link | Pretrained Models|Static Models|Size (static)
Model Type | Dataset| Example Link | Pretrained Models|Static/ONNX Models|Size (static)
:-------------:| :------------:| :-----: | :-----:| :-----:| :-----:
Tacotron2|LJSpeech|[tacotron2-ljspeech](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/tts0)|[tacotron2_ljspeech_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/tacotron2/tacotron2_ljspeech_ckpt_0.2.0.zip)|||
Tacotron2|CSMSC|[tacotron2-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/tts0)|[tacotron2_csmsc_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/tacotron2/tacotron2_csmsc_ckpt_0.2.0.zip)|[tacotron2_csmsc_static_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/tacotron2/tacotron2_csmsc_static_0.2.0.zip)|103MB|
TransformerTTS| LJSpeech| [transformer-ljspeech](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/tts1)|[transformer_tts_ljspeech_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/transformer_tts/transformer_tts_ljspeech_ckpt_0.4.zip)|||
SpeedySpeech| CSMSC | [speedyspeech-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/tts2)|[speedyspeech_csmsc_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/speedyspeech/speedyspeech_csmsc_ckpt_0.2.0.zip)|[speedyspeech_csmsc_static_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/speedyspeech/speedyspeech_csmsc_static_0.2.0.zip)|12MB|
FastSpeech2| CSMSC |[fastspeech2-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/tts3)|[fastspeech2_nosil_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_nosil_baker_ckpt_0.4.zip)|[fastspeech2_csmsc_static_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_csmsc_static_0.2.0.zip)|157MB|
SpeedySpeech| CSMSC | [speedyspeech-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/tts2)|[speedyspeech_csmsc_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/speedyspeech/speedyspeech_csmsc_ckpt_0.2.0.zip)|[speedyspeech_csmsc_static_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/speedyspeech/speedyspeech_csmsc_static_0.2.0.zip) </br> [speedyspeech_csmsc_onnx_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/speedyspeech/speedyspeech_csmsc_onnx_0.2.0.zip)|13MB|
FastSpeech2| CSMSC |[fastspeech2-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/tts3)|[fastspeech2_nosil_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_nosil_baker_ckpt_0.4.zip)|[fastspeech2_csmsc_static_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_csmsc_static_0.2.0.zip) </br> [fastspeech2_csmsc_onnx_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_csmsc_onnx_0.2.0.zip)|157MB|
FastSpeech2-Conformer| CSMSC |[fastspeech2-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/tts3)|[fastspeech2_conformer_baker_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_conformer_baker_ckpt_0.5.zip)|||
FastSpeech2| AISHELL-3 |[fastspeech2-aishell3](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/tts3)|[fastspeech2_nosil_aishell3_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_nosil_aishell3_ckpt_0.4.zip)|||
FastSpeech2| LJSpeech |[fastspeech2-ljspeech](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/tts3)|[fastspeech2_nosil_ljspeech_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_nosil_ljspeech_ckpt_0.5.zip)|||
FastSpeech2| VCTK |[fastspeech2-vctk](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/vctk/tts3)|[fastspeech2_nosil_vctk_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_nosil_vctk_ckpt_0.5.zip)|||
FastSpeech2-CNNDecoder| CSMSC| [fastspeech2-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/tts3)| [fastspeech2_cnndecoder_csmsc_ckpt_1.0.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_cnndecoder_csmsc_ckpt_1.0.0.zip) | [fastspeech2_cnndecoder_csmsc_static_1.0.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_cnndecoder_csmsc_static_1.0.0.zip) </br>[fastspeech2_cnndecoder_csmsc_streaming_static_1.0.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_cnndecoder_csmsc_streaming_static_1.0.0.zip) </br>[fastspeech2_cnndecoder_csmsc_onnx_1.0.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_cnndecoder_csmsc_onnx_1.0.0.zip) </br>[fastspeech2_cnndecoder_csmsc_streaming_onnx_1.0.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_cnndecoder_csmsc_streaming_onnx_1.0.0.zip) | 84MB|
FastSpeech2| AISHELL-3 |[fastspeech2-aishell3](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/tts3)|[fastspeech2_nosil_aishell3_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_nosil_aishell3_ckpt_0.4.zip)|[fastspeech2_aishell3_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_aishell3_static_1.1.0.zip) </br> [fastspeech2_aishell3_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_aishell3_onnx_1.1.0.zip)|147MB|
FastSpeech2| LJSpeech |[fastspeech2-ljspeech](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/tts3)|[fastspeech2_nosil_ljspeech_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_nosil_ljspeech_ckpt_0.5.zip)|[fastspeech2_ljspeech_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_ljspeech_static_1.1.0.zip) </br> [fastspeech2_ljspeech_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_ljspeech_onnx_1.1.0.zip)|145MB|
FastSpeech2| VCTK |[fastspeech2-vctk](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/vctk/tts3)|[fastspeech2_nosil_vctk_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_nosil_vctk_ckpt_0.5.zip)|[fastspeech2_vctk_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_vctk_static_1.1.0.zip) </br> [fastspeech2_vctk_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_vctk_onnx_1.1.0.zip) | 145MB|
### Vocoders
Model Type | Dataset| Example Link | Pretrained Models| Static Models|Size (static)
Model Type | Dataset| Example Link | Pretrained Models| Static/ONNX Models|Size (static)
:-----:| :-----:| :-----: | :-----:| :-----:| :-----:
WaveFlow| LJSpeech |[waveflow-ljspeech](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc0)|[waveflow_ljspeech_ckpt_0.3.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/waveflow/waveflow_ljspeech_ckpt_0.3.zip)|||
Parallel WaveGAN| CSMSC |[PWGAN-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/voc1)|[pwg_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwg_baker_ckpt_0.4.zip)|[pwg_baker_static_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwg_baker_static_0.4.zip)|5.1MB|
Parallel WaveGAN| LJSpeech |[PWGAN-ljspeech](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc1)|[pwg_ljspeech_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwg_ljspeech_ckpt_0.5.zip)|||
Parallel WaveGAN| AISHELL-3 |[PWGAN-aishell3](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/voc1)|[pwg_aishell3_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwg_aishell3_ckpt_0.5.zip)|||
Parallel WaveGAN| VCTK |[PWGAN-vctk](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/vctk/voc1)|[pwg_vctk_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwg_vctk_ckpt_0.5.zip)|||
|Multi Band MelGAN | CSMSC |[MB MelGAN-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/voc3) | [mb_melgan_csmsc_ckpt_0.1.1.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/mb_melgan/mb_melgan_csmsc_ckpt_0.1.1.zip) <br>[mb_melgan_baker_finetune_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/mb_melgan/mb_melgan_baker_finetune_ckpt_0.5.zip)|[mb_melgan_csmsc_static_0.1.1.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/mb_melgan/mb_melgan_csmsc_static_0.1.1.zip) |8.2MB|
Parallel WaveGAN| CSMSC |[PWGAN-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/voc1)|[pwg_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwg_baker_ckpt_0.4.zip)|[pwg_baker_static_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwg_baker_static_0.4.zip) </br> [pwgan_csmsc_onnx_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwgan_csmsc_onnx_0.2.0.zip)|4.8MB|
Parallel WaveGAN| LJSpeech |[PWGAN-ljspeech](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc1)|[pwg_ljspeech_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwg_ljspeech_ckpt_0.5.zip)|[pwgan_ljspeech_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwgan_ljspeech_static_1.1.0.zip) </br> [pwgan_ljspeech_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwgan_ljspeech_onnx_1.1.0.zip)|4.8MB|
Parallel WaveGAN| AISHELL-3 |[PWGAN-aishell3](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/voc1)|[pwg_aishell3_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwg_aishell3_ckpt_0.5.zip)| [pwgan_aishell3_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwgan_aishell3_static_1.1.0.zip) </br> [pwgan_aishell3_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwgan_aishell3_onnx_1.1.0.zip)|4.8MB|
Parallel WaveGAN| VCTK |[PWGAN-vctk](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/vctk/voc1)|[pwg_vctk_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwg_vctk_ckpt_0.5.zip)|[pwgan_vctk_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwgan_vctk_static_1.1.0.zip) </br> [pwgan_vctk_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwgan_vctk_onnx_1.1.0.zip)|4.8MB|
|Multi Band MelGAN | CSMSC |[MB MelGAN-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/voc3) | [mb_melgan_csmsc_ckpt_0.1.1.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/mb_melgan/mb_melgan_csmsc_ckpt_0.1.1.zip) <br>[mb_melgan_baker_finetune_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/mb_melgan/mb_melgan_baker_finetune_ckpt_0.5.zip)|[mb_melgan_csmsc_static_0.1.1.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/mb_melgan/mb_melgan_csmsc_static_0.1.1.zip) </br> [mb_melgan_csmsc_onnx_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/mb_melgan/mb_melgan_csmsc_onnx_0.2.0.zip)|7.6MB|
Style MelGAN | CSMSC |[Style MelGAN-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/voc4)|[style_melgan_csmsc_ckpt_0.1.1.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/style_melgan/style_melgan_csmsc_ckpt_0.1.1.zip)| | |
HiFiGAN | CSMSC |[HiFiGAN-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/voc5)|[hifigan_csmsc_ckpt_0.1.1.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_csmsc_ckpt_0.1.1.zip)|[hifigan_csmsc_static_0.1.1.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_csmsc_static_0.1.1.zip)|50MB|
HiFiGAN | LJSpeech |[HiFiGAN-ljspeech](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc5)|[hifigan_ljspeech_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_ljspeech_ckpt_0.2.0.zip)|||
HiFiGAN | AISHELL-3 |[HiFiGAN-aishell3](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/voc5)|[hifigan_aishell3_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_aishell3_ckpt_0.2.0.zip)|||
HiFiGAN | VCTK |[HiFiGAN-vctk](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/vctk/voc5)|[hifigan_vctk_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_vctk_ckpt_0.2.0.zip)|||
HiFiGAN | CSMSC |[HiFiGAN-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/voc5)|[hifigan_csmsc_ckpt_0.1.1.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_csmsc_ckpt_0.1.1.zip)|[hifigan_csmsc_static_0.1.1.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_csmsc_static_0.1.1.zip) </br> [hifigan_csmsc_onnx_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_csmsc_onnx_0.2.0.zip)|46MB|
HiFiGAN | LJSpeech |[HiFiGAN-ljspeech](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc5)|[hifigan_ljspeech_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_ljspeech_ckpt_0.2.0.zip)|[hifigan_ljspeech_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_ljspeech_static_1.1.0.zip) </br> [hifigan_ljspeech_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_ljspeech_onnx_1.1.0.zip) |49MB|
HiFiGAN | AISHELL-3 |[HiFiGAN-aishell3](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/voc5)|[hifigan_aishell3_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_aishell3_ckpt_0.2.0.zip)|[hifigan_aishell3_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_aishell3_static_1.1.0.zip) </br> [hifigan_aishell3_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_aishell3_onnx_1.1.0.zip)|46MB|
HiFiGAN | VCTK |[HiFiGAN-vctk](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/vctk/voc5)|[hifigan_vctk_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_vctk_ckpt_0.2.0.zip)|[hifigan_vctk_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_vctk_static_1.1.0.zip) </br> [hifigan_vctk_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_vctk_onnx_1.1.0.zip)|46MB|
WaveRNN | CSMSC |[WaveRNN-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/voc6)|[wavernn_csmsc_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/wavernn/wavernn_csmsc_ckpt_0.2.0.zip)|[wavernn_csmsc_static_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/wavernn/wavernn_csmsc_static_0.2.0.zip)|18MB|
......
examples/aishell3/tts3/README.md
浏览文件 @ 0fa3fdb9
......@@ -220,6 +220,12 @@ Pretrained FastSpeech2 model with no silence in the edge of audios:
- [fastspeech2_nosil_aishell3_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_nosil_aishell3_ckpt_0.4.zip)
- [fastspeech2_conformer_aishell3_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_conformer_aishell3_ckpt_0.2.0.zip) (Thanks for [@awmmmm](https://github.com/awmmmm)'s contribution)
The static model can be downloaded here:
- [fastspeech2_aishell3_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_aishell3_static_1.1.0.zip)
The ONNX model can be downloaded here:
- [fastspeech2_aishell3_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_aishell3_onnx_1.1.0.zip)
FastSpeech2 checkpoint contains files listed below.
```text
......
examples/aishell3/tts3/local/inference.sh
浏览文件 @ 0fa3fdb9
......@@ -17,3 +17,14 @@ if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
--spk_id=0
fi
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
python3 ${BIN_DIR}/../inference.py \
--inference_dir=${train_output_path}/inference \
--am=fastspeech2_aishell3 \
--voc=hifigan_aishell3 \
--text=${BIN_DIR}/../sentences.txt \
--output_dir=${train_output_path}/pd_infer_out \
--phones_dict=dump/phone_id_map.txt \
--speaker_dict=dump/speaker_id_map.txt \
--spk_id=0
fi
examples/aishell3/tts3/local/ort_predict.sh 0 → 100755
浏览文件 @ 0fa3fdb9
train_output_path=$1
stage=0
stop_stage=0
# e2e, synthesize from text
if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
python3 ${BIN_DIR}/../ort_predict_e2e.py \
--inference_dir=${train_output_path}/inference_onnx \
--am=fastspeech2_aishell3 \
--voc=pwgan_aishell3 \
--output_dir=${train_output_path}/onnx_infer_out_e2e \
--text=${BIN_DIR}/../csmsc_test.txt \
--phones_dict=dump/phone_id_map.txt \
--device=cpu \
--cpu_threads=2 \
--spk_id=0
fi
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
python3 ${BIN_DIR}/../ort_predict_e2e.py \
--inference_dir=${train_output_path}/inference_onnx \
--am=fastspeech2_aishell3 \
--voc=hifigan_aishell3 \
--output_dir=${train_output_path}/onnx_infer_out_e2e \
--text=${BIN_DIR}/../csmsc_test.txt \
--phones_dict=dump/phone_id_map.txt \
--device=cpu \
--cpu_threads=2 \
--spk_id=0
fi
examples/aishell3/tts3/local/paddle2onnx.sh 0 → 120000
浏览文件 @ 0fa3fdb9
../../../csmsc/tts3/local/paddle2onnx.sh
\ No newline at end of file
examples/aishell3/tts3/run.sh
浏览文件 @ 0fa3fdb9
......@@ -27,11 +27,34 @@ if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
fi
if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
# synthesize, vocoder is pwgan
# synthesize, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize.sh ${conf_path} ${train_output_path} ${ckpt_name} || exit -1
fi
if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
# synthesize_e2e, vocoder is pwgan
# synthesize_e2e, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize_e2e.sh ${conf_path} ${train_output_path} ${ckpt_name} || exit -1
fi
if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
# inference with static model, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/inference.sh ${train_output_path} || exit -1
fi
if [ ${stage} -le 5 ] && [ ${stop_stage} -ge 5 ]; then
# install paddle2onnx
version=$(echo `pip list |grep "paddle2onnx"` |awk -F" " '{print $2}')
if [[ -z "$version" || ${version} != '0.9.8' ]]; then
pip install paddle2onnx==0.9.8
fi
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx fastspeech2_aishell3
# considering the balance between speed and quality, we recommend that you use hifigan as vocoder
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx pwgan_aishell3
# ./local/paddle2onnx.sh ${train_output_path} inference inference_onnx hifigan_aishell3
fi
# inference with onnxruntime, use fastspeech2 + hifigan by default
if [ ${stage} -le 6 ] && [ ${stop_stage} -ge 6 ]; then
./local/ort_predict.sh ${train_output_path}
fi
examples/aishell3/voc1/README.md
浏览文件 @ 0fa3fdb9
......@@ -133,6 +133,12 @@ optional arguments:
Pretrained models can be downloaded here:
- [pwg_aishell3_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwg_aishell3_ckpt_0.5.zip)
The static model can be downloaded here:
- [pwgan_aishell3_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwgan_aishell3_static_1.1.0.zip)
The ONNX model can be downloaded here:
- [pwgan_aishell3_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwgan_aishell3_onnx_1.1.0.zip)
Model | Step | eval/generator_loss | eval/log_stft_magnitude_loss:| eval/spectral_convergence_loss
:-------------:| :------------:| :-----: | :-----: | :--------:
default| 1(gpu) x 400000|1.968762|0.759008|0.218524
......
examples/aishell3/voc5/README.md
浏览文件 @ 0fa3fdb9
......@@ -116,6 +116,11 @@ optional arguments:
The pretrained model can be downloaded here:
- [hifigan_aishell3_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_aishell3_ckpt_0.2.0.zip)
The static model can be downloaded here:
- [hifigan_aishell3_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_aishell3_static_1.1.0.zip)
The ONNX model can be downloaded here:
- [hifigan_aishell3_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_aishell3_onnx_1.1.0.zip)
Model | Step | eval/generator_loss | eval/mel_loss| eval/feature_matching_loss
:-------------:| :------------:| :-----: | :-----: | :--------:
......
examples/csmsc/tts2/local/ort_predict.sh
浏览文件 @ 0fa3fdb9
......@@ -3,22 +3,34 @@ train_output_path=$1
stage=0
stop_stage=0
# only support default_fastspeech2/speedyspeech + hifigan/mb_melgan now!
# synthesize from metadata
# e2e, synthesize from text
if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
python3 ${BIN_DIR}/../ort_predict.py \
python3 ${BIN_DIR}/../ort_predict_e2e.py \
--inference_dir=${train_output_path}/inference_onnx \
--am=speedyspeech_csmsc \
--voc=hifigan_csmsc \
--test_metadata=dump/test/norm/metadata.jsonl \
--output_dir=${train_output_path}/onnx_infer_out \
--voc=pwgan_csmsc \
--output_dir=${train_output_path}/onnx_infer_out_e2e \
--text=${BIN_DIR}/../csmsc_test.txt \
--phones_dict=dump/phone_id_map.txt \
--tones_dict=dump/tone_id_map.txt \
--device=cpu \
--cpu_threads=2
fi
# e2e, synthesize from text
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
python3 ${BIN_DIR}/../ort_predict_e2e.py \
--inference_dir=${train_output_path}/inference_onnx \
--am=speedyspeech_csmsc \
--voc=mb_melgan_csmsc \
--output_dir=${train_output_path}/onnx_infer_out_e2e \
--text=${BIN_DIR}/../csmsc_test.txt \
--phones_dict=dump/phone_id_map.txt \
--tones_dict=dump/tone_id_map.txt \
--device=cpu \
--cpu_threads=2
fi
if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
python3 ${BIN_DIR}/../ort_predict_e2e.py \
--inference_dir=${train_output_path}/inference_onnx \
--am=speedyspeech_csmsc \
......@@ -30,3 +42,15 @@ if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
--device=cpu \
--cpu_threads=2
fi
# synthesize from metadata
if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
python3 ${BIN_DIR}/../ort_predict.py \
--inference_dir=${train_output_path}/inference_onnx \
--am=speedyspeech_csmsc \
--voc=hifigan_csmsc \
--test_metadata=dump/test/norm/metadata.jsonl \
--output_dir=${train_output_path}/onnx_infer_out \
--device=cpu \
--cpu_threads=2
fi
examples/csmsc/tts2/run.sh
浏览文件 @ 0fa3fdb9
......@@ -27,12 +27,12 @@ if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
fi
if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
# synthesize, vocoder is pwgan
# synthesize, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize.sh ${conf_path} ${train_output_path} ${ckpt_name} || exit -1
fi
if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
# synthesize_e2e, vocoder is pwgan
# synthesize_e2e, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize_e2e.sh ${conf_path} ${train_output_path} ${ckpt_name} || exit -1
fi
......@@ -46,19 +46,17 @@ fi
if [ ${stage} -le 5 ] && [ ${stop_stage} -ge 5 ]; then
# install paddle2onnx
version=$(echo `pip list |grep "paddle2onnx"` |awk -F" " '{print $2}')
if [[ -z "$version" || ${version} != '0.9.5' ]]; then
pip install paddle2onnx==0.9.5
if [[ -z "$version" || ${version} != '0.9.8' ]]; then
pip install paddle2onnx==0.9.8
fi
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx speedyspeech_csmsc
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx hifigan_csmsc
# considering the balance between speed and quality, we recommend that you use hifigan as vocoder
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx pwgan_csmsc
# ./local/paddle2onnx.sh ${train_output_path} inference inference_onnx mb_melgan_csmsc
# ./local/paddle2onnx.sh ${train_output_path} inference inference_onnx hifigan_csmsc
fi
# inference with onnxruntime, use fastspeech2 + hifigan by default
# inference with onnxruntime
if [ ${stage} -le 6 ] && [ ${stop_stage} -ge 6 ]; then
# install onnxruntime
version=$(echo `pip list |grep "onnxruntime"` |awk -F" " '{print $2}')
if [[ -z "$version" || ${version} != '1.10.0' ]]; then
pip install onnxruntime==1.10.0
fi
./local/ort_predict.sh ${train_output_path}
fi
examples/csmsc/tts3/local/ort_predict.sh
浏览文件 @ 0fa3fdb9
......@@ -3,22 +3,32 @@ train_output_path=$1
stage=0
stop_stage=0
# only support default_fastspeech2/speedyspeech + hifigan/mb_melgan now!
# synthesize from metadata
# e2e, synthesize from text
if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
python3 ${BIN_DIR}/../ort_predict.py \
python3 ${BIN_DIR}/../ort_predict_e2e.py \
--inference_dir=${train_output_path}/inference_onnx \
--am=fastspeech2_csmsc \
--voc=hifigan_csmsc \
--test_metadata=dump/test/norm/metadata.jsonl \
--output_dir=${train_output_path}/onnx_infer_out \
--voc=pwgan_csmsc \
--output_dir=${train_output_path}/onnx_infer_out_e2e \
--text=${BIN_DIR}/../csmsc_test.txt \
--phones_dict=dump/phone_id_map.txt \
--device=cpu \
--cpu_threads=2
fi
# e2e, synthesize from text
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
python3 ${BIN_DIR}/../ort_predict_e2e.py \
--inference_dir=${train_output_path}/inference_onnx \
--am=fastspeech2_csmsc \
--voc=mb_melgan_csmsc \
--output_dir=${train_output_path}/onnx_infer_out_e2e \
--text=${BIN_DIR}/../csmsc_test.txt \
--phones_dict=dump/phone_id_map.txt \
--device=cpu \
--cpu_threads=2
fi
if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
python3 ${BIN_DIR}/../ort_predict_e2e.py \
--inference_dir=${train_output_path}/inference_onnx \
--am=fastspeech2_csmsc \
......@@ -29,3 +39,15 @@ if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
--device=cpu \
--cpu_threads=2
fi
# synthesize from metadata, take hifigan as an example
if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
python3 ${BIN_DIR}/../ort_predict.py \
--inference_dir=${train_output_path}/inference_onnx \
--am=fastspeech2_csmsc \
--voc=hifigan_csmsc \
--test_metadata=dump/test/norm/metadata.jsonl \
--output_dir=${train_output_path}/onnx_infer_out \
--device=cpu \
--cpu_threads=2
fi
\ No newline at end of file
examples/csmsc/tts3/local/ort_predict_streaming.sh
浏览文件 @ 0fa3fdb9
......@@ -5,6 +5,34 @@ stop_stage=0
# e2e, synthesize from text
if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
python3 ${BIN_DIR}/../ort_predict_streaming.py \
--inference_dir=${train_output_path}/inference_onnx_streaming \
--am=fastspeech2_csmsc \
--am_stat=dump/train/speech_stats.npy \
--voc=pwgan_csmsc \
--output_dir=${train_output_path}/onnx_infer_out_streaming \
--text=${BIN_DIR}/../csmsc_test.txt \
--phones_dict=dump/phone_id_map.txt \
--device=cpu \
--cpu_threads=2 \
--am_streaming=True
fi
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
python3 ${BIN_DIR}/../ort_predict_streaming.py \
--inference_dir=${train_output_path}/inference_onnx_streaming \
--am=fastspeech2_csmsc \
--am_stat=dump/train/speech_stats.npy \
--voc=mb_melgan_csmsc \
--output_dir=${train_output_path}/onnx_infer_out_streaming \
--text=${BIN_DIR}/../csmsc_test.txt \
--phones_dict=dump/phone_id_map.txt \
--device=cpu \
--cpu_threads=2 \
--am_streaming=True
fi
if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
python3 ${BIN_DIR}/../ort_predict_streaming.py \
--inference_dir=${train_output_path}/inference_onnx_streaming \
--am=fastspeech2_csmsc \
......
examples/csmsc/tts3/local/synthesize_streaming.sh
浏览文件 @ 0fa3fdb9
......@@ -24,7 +24,8 @@ if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
--text=${BIN_DIR}/../sentences.txt \
--output_dir=${train_output_path}/test_e2e_streaming \
--phones_dict=dump/phone_id_map.txt \
--am_streaming=True
--am_streaming=True \
--inference_dir=${train_output_path}/inference_streaming
fi
# for more GAN Vocoders
......@@ -45,7 +46,8 @@ if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
--text=${BIN_DIR}/../sentences.txt \
--output_dir=${train_output_path}/test_e2e_streaming \
--phones_dict=dump/phone_id_map.txt \
--am_streaming=True
--am_streaming=True \
--inference_dir=${train_output_path}/inference_streaming
fi
# the pretrained models haven't release now
......
examples/csmsc/tts3/run.sh
浏览文件 @ 0fa3fdb9
......@@ -27,17 +27,17 @@ if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
fi
if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
# synthesize, vocoder is pwgan
# synthesize, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize.sh ${conf_path} ${train_output_path} ${ckpt_name} || exit -1
fi
if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
# synthesize_e2e, vocoder is pwgan
# synthesize_e2e, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize_e2e.sh ${conf_path} ${train_output_path} ${ckpt_name} || exit -1
fi
if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
# inference with static model
# inference with static model, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/inference.sh ${train_output_path} || exit -1
fi
......@@ -46,15 +46,18 @@ fi
if [ ${stage} -le 5 ] && [ ${stop_stage} -ge 5 ]; then
# install paddle2onnx
version=$(echo `pip list |grep "paddle2onnx"` |awk -F" " '{print $2}')
if [[ -z "$version" || ${version} != '0.9.5' ]]; then
pip install paddle2onnx==0.9.5
if [[ -z "$version" || ${version} != '0.9.8' ]]; then
pip install paddle2onnx==0.9.8
fi
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx fastspeech2_csmsc
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx hifigan_csmsc
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx mb_melgan_csmsc
# considering the balance between speed and quality, we recommend that you use hifigan as vocoder
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx pwgan_csmsc
# ./local/paddle2onnx.sh ${train_output_path} inference inference_onnx mb_melgan_csmsc
# ./local/paddle2onnx.sh ${train_output_path} inference inference_onnx hifigan_csmsc
fi
# inference with onnxruntime, use fastspeech2 + hifigan by default
# inference with onnxruntime
if [ ${stage} -le 6 ] && [ ${stop_stage} -ge 6 ]; then
./local/ort_predict.sh ${train_output_path}
fi
examples/csmsc/tts3/run_cnndecoder.sh
浏览文件 @ 0fa3fdb9
......@@ -33,25 +33,25 @@ fi
# synthesize_e2e non-streaming
if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
# synthesize_e2e, vocoder is pwgan
# synthesize_e2e, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize_e2e.sh ${conf_path} ${train_output_path} ${ckpt_name} || exit -1
fi
# inference non-streaming
if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
# inference with static model
# inference with static model, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/inference.sh ${train_output_path} || exit -1
fi
# synthesize_e2e streaming
if [ ${stage} -le 5 ] && [ ${stop_stage} -ge 5 ]; then
# synthesize_e2e, vocoder is pwgan
# synthesize_e2e, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize_streaming.sh ${conf_path} ${train_output_path} ${ckpt_name} || exit -1
fi
# inference streaming
if [ ${stage} -le 6 ] && [ ${stop_stage} -ge 6 ]; then
# inference with static model
# inference with static model, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/inference_streaming.sh ${train_output_path} || exit -1
fi
......@@ -59,32 +59,37 @@ fi
if [ ${stage} -le 7 ] && [ ${stop_stage} -ge 7 ]; then
# install paddle2onnx
version=$(echo `pip list |grep "paddle2onnx"` |awk -F" " '{print $2}')
if [[ -z "$version" || ${version} != '0.9.5' ]]; then
pip install paddle2onnx==0.9.5
if [[ -z "$version" || ${version} != '0.9.8' ]]; then
pip install paddle2onnx==0.9.8
fi
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx fastspeech2_csmsc
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx hifigan_csmsc
# considering the balance between speed and quality, we recommend that you use hifigan as vocoder
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx pwgan_csmsc
# ./local/paddle2onnx.sh ${train_output_path} inference inference_onnx mb_melgan_csmsc
# ./local/paddle2onnx.sh ${train_output_path} inference inference_onnx hifigan_csmsc
fi
# onnxruntime non streaming
# inference with onnxruntime, use fastspeech2 + hifigan by default
if [ ${stage} -le 8 ] && [ ${stop_stage} -ge 8 ]; then
./local/ort_predict.sh ${train_output_path}
fi
# paddle2onnx streaming
if [ ${stage} -le 9 ] && [ ${stop_stage} -ge 9 ]; then
# install paddle2onnx
version=$(echo `pip list |grep "paddle2onnx"` |awk -F" " '{print $2}')
if [[ -z "$version" || ${version} != '0.9.5' ]]; then
pip install paddle2onnx==0.9.5
if [[ -z "$version" || ${version} != '0.9.8' ]]; then
pip install paddle2onnx==0.9.8
fi
# streaming acoustic model
./local/paddle2onnx.sh ${train_output_path} inference_streaming inference_onnx_streaming fastspeech2_csmsc_am_encoder_infer
./local/paddle2onnx.sh ${train_output_path} inference_streaming inference_onnx_streaming fastspeech2_csmsc_am_decoder
./local/paddle2onnx.sh ${train_output_path} inference_streaming inference_onnx_streaming fastspeech2_csmsc_am_postnet
# vocoder
./local/paddle2onnx.sh ${train_output_path} inference_streaming inference_onnx_streaming hifigan_csmsc
# considering the balance between speed and quality, we recommend that you use hifigan as vocoder
./local/paddle2onnx.sh ${train_output_path} inference_streaming inference_onnx_streaming pwgan_csmsc
# ./local/paddle2onnx.sh ${train_output_path} inference_streaming inference_onnx_streaming mb_melgan_csmsc
# ./local/paddle2onnx.sh ${train_output_path} inference_streaming inference_onnx_streaming hifigan_csmsc
fi
# onnxruntime streaming
......
examples/ljspeech/tts3/README.md
浏览文件 @ 0fa3fdb9
......@@ -215,6 +215,13 @@ optional arguments:
Pretrained FastSpeech2 model with no silence in the edge of audios:
- [fastspeech2_nosil_ljspeech_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_nosil_ljspeech_ckpt_0.5.zip)
The static model can be downloaded here:
- [fastspeech2_ljspeech_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_ljspeech_static_1.1.0.zip)
The ONNX model can be downloaded here:
- [fastspeech2_ljspeech_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_ljspeech_onnx_1.1.0.zip)
Model | Step | eval/loss | eval/l1_loss | eval/duration_loss | eval/pitch_loss| eval/energy_loss
:-------------:| :------------:| :-----: | :-----: | :--------: |:--------:|:---------:
default| 2(gpu) x 100000| 1.505682|0.612104| 0.045505| 0.62792| 0.220147
......
examples/ljspeech/tts3/local/inference.sh 0 → 100755
浏览文件 @ 0fa3fdb9
#!/bin/bash
train_output_path=$1
stage=0
stop_stage=0
# pwgan
if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
python3 ${BIN_DIR}/../inference.py \
--inference_dir=${train_output_path}/inference \
--am=fastspeech2_ljspeech \
--voc=pwgan_ljspeech \
--text=${BIN_DIR}/../sentences_en.txt \
--output_dir=${train_output_path}/pd_infer_out \
--phones_dict=dump/phone_id_map.txt \
--lang=en
fi
# hifigan
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
python3 ${BIN_DIR}/../inference.py \
--inference_dir=${train_output_path}/inference \
--am=fastspeech2_ljspeech \
--voc=hifigan_ljspeech \
--text=${BIN_DIR}/../sentences_en.txt \
--output_dir=${train_output_path}/pd_infer_out \
--phones_dict=dump/phone_id_map.txt \
--lang=en
fi
examples/ljspeech/tts3/local/ort_predict.sh 0 → 100755
浏览文件 @ 0fa3fdb9
train_output_path=$1
stage=0
stop_stage=0
# e2e, synthesize from text
if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
python3 ${BIN_DIR}/../ort_predict_e2e.py \
--inference_dir=${train_output_path}/inference_onnx \
--am=fastspeech2_ljspeech \
--voc=pwgan_ljspeech\
--output_dir=${train_output_path}/onnx_infer_out_e2e \
--text=${BIN_DIR}/../sentences_en.txt \
--phones_dict=dump/phone_id_map.txt \
--device=cpu \
--cpu_threads=2 \
--lang=en
fi
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
python3 ${BIN_DIR}/../ort_predict_e2e.py \
--inference_dir=${train_output_path}/inference_onnx \
--am=fastspeech2_ljspeech \
--voc=hifigan_ljspeech \
--output_dir=${train_output_path}/onnx_infer_out_e2e \
--text=${BIN_DIR}/../sentences_en.txt \
--phones_dict=dump/phone_id_map.txt \
--device=cpu \
--cpu_threads=2 \
--lang=en
fi
examples/ljspeech/tts3/local/paddle2onnx.sh 0 → 120000
浏览文件 @ 0fa3fdb9
../../../csmsc/tts3/local/paddle2onnx.sh
\ No newline at end of file
examples/ljspeech/tts3/run.sh
浏览文件 @ 0fa3fdb9
......@@ -27,11 +27,35 @@ if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
fi
if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
# synthesize, vocoder is pwgan
# synthesize, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize.sh ${conf_path} ${train_output_path} ${ckpt_name} || exit -1
fi
if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
# synthesize_e2e, vocoder is pwgan
# synthesize_e2e, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize_e2e.sh ${conf_path} ${train_output_path} ${ckpt_name} || exit -1
fi
if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
# inference with static model, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/inference.sh ${train_output_path} || exit -1
fi
# paddle2onnx, please make sure the static models are in ${train_output_path}/inference first
# we have only tested the following models so far
if [ ${stage} -le 5 ] && [ ${stop_stage} -ge 5 ]; then
# install paddle2onnx
version=$(echo `pip list |grep "paddle2onnx"` |awk -F" " '{print $2}')
if [[ -z "$version" || ${version} != '0.9.8' ]]; then
pip install paddle2onnx==0.9.8
fi
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx fastspeech2_ljspeech
# considering the balance between speed and quality, we recommend that you use hifigan as vocoder
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx pwgan_ljspeech
# ./local/paddle2onnx.sh ${train_output_path} inference inference_onnx hifigan_ljspeech
fi
# inference with onnxruntime, use fastspeech2 + hifigan by default
if [ ${stage} -le 6 ] && [ ${stop_stage} -ge 6 ]; then
./local/ort_predict.sh ${train_output_path}
fi
examples/ljspeech/voc1/README.md
浏览文件 @ 0fa3fdb9
......@@ -130,6 +130,13 @@ optional arguments:
Pretrained models can be downloaded here:
- [pwg_ljspeech_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwg_ljspeech_ckpt_0.5.zip)
The static model can be downloaded here:
- [pwgan_ljspeech_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwgan_ljspeech_static_1.1.0.zip)
The ONNX model can be downloaded here:
- [pwgan_ljspeech_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwgan_ljspeech_onnx_1.1.0.zip)
Parallel WaveGAN checkpoint contains files listed below.
```text
......
examples/ljspeech/voc5/README.md
浏览文件 @ 0fa3fdb9
......@@ -115,6 +115,12 @@ optional arguments:
The pretrained model can be downloaded here:
- [hifigan_ljspeech_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_ljspeech_ckpt_0.2.0.zip)
The static model can be downloaded here:
- [hifigan_ljspeech_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_ljspeech_static_1.1.0.zip)
The ONNX model can be downloaded here:
- [hifigan_ljspeech_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_ljspeech_onnx_1.1.0.zip)
Model | Step | eval/generator_loss | eval/mel_loss| eval/feature_matching_loss
:-------------:| :------------:| :-----: | :-----: | :--------:
......
examples/vctk/tts3/README.md
浏览文件 @ 0fa3fdb9
......@@ -218,6 +218,12 @@ optional arguments:
Pretrained FastSpeech2 model with no silence in the edge of audios:
- [fastspeech2_nosil_vctk_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_nosil_vctk_ckpt_0.5.zip)
The static model can be downloaded here:
- [fastspeech2_vctk_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_vctk_static_1.1.0.zip)
The ONNX model can be downloaded here:
- [fastspeech2_vctk_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/fastspeech2/fastspeech2_vctk_onnx_1.1.0.zip)
FastSpeech2 checkpoint contains files listed below.
```text
fastspeech2_nosil_vctk_ckpt_0.5
......
examples/vctk/tts3/local/inference.sh
浏览文件 @ 0fa3fdb9
......@@ -18,3 +18,15 @@ if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
--lang=en
fi
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
python3 ${BIN_DIR}/../inference.py \
--inference_dir=${train_output_path}/inference \
--am=fastspeech2_vctk \
--voc=hifigan_vctk \
--text=${BIN_DIR}/../sentences_en.txt \
--output_dir=${train_output_path}/pd_infer_out \
--phones_dict=dump/phone_id_map.txt \
--speaker_dict=dump/speaker_id_map.txt \
--spk_id=0 \
--lang=en
fi
examples/vctk/tts3/local/ort_predict.sh 0 → 100755
浏览文件 @ 0fa3fdb9
train_output_path=$1
stage=0
stop_stage=0
# e2e, synthesize from text
if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
python3 ${BIN_DIR}/../ort_predict_e2e.py \
--inference_dir=${train_output_path}/inference_onnx \
--am=fastspeech2_vctk \
--voc=pwgan_vctk \
--output_dir=${train_output_path}/onnx_infer_out_e2e \
--text=${BIN_DIR}/../sentences_en.txt \
--phones_dict=dump/phone_id_map.txt \
--device=cpu \
--cpu_threads=2 \
--spk_id=0 \
--lang=en
fi
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
python3 ${BIN_DIR}/../ort_predict_e2e.py \
--inference_dir=${train_output_path}/inference_onnx \
--am=fastspeech2_vctk \
--voc=hifigan_vctk \
--output_dir=${train_output_path}/onnx_infer_out_e2e \
--text=${BIN_DIR}/../sentences_en.txt \
--phones_dict=dump/phone_id_map.txt \
--device=cpu \
--cpu_threads=2 \
--spk_id=0 \
--lang=en
fi
examples/vctk/tts3/local/paddle2onnx.sh 0 → 120000
浏览文件 @ 0fa3fdb9
../../../csmsc/tts3/local/paddle2onnx.sh
\ No newline at end of file
examples/vctk/tts3/run.sh
浏览文件 @ 0fa3fdb9
......@@ -27,11 +27,34 @@ if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
fi
if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
# synthesize, vocoder is pwgan
# synthesize, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize.sh ${conf_path} ${train_output_path} ${ckpt_name} || exit -1
fi
if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
# synthesize_e2e, vocoder is pwgan
# synthesize_e2e, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize_e2e.sh ${conf_path} ${train_output_path} ${ckpt_name} || exit -1
fi
if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
# inference with static model, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/inference.sh ${train_output_path} || exit -1
fi
if [ ${stage} -le 5 ] && [ ${stop_stage} -ge 5 ]; then
# install paddle2onnx
version=$(echo `pip list |grep "paddle2onnx"` |awk -F" " '{print $2}')
if [[ -z "$version" || ${version} != '0.9.8' ]]; then
pip install paddle2onnx==0.9.8
fi
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx fastspeech2_vctk
# considering the balance between speed and quality, we recommend that you use hifigan as vocoder
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx pwgan_vctk
# ./local/paddle2onnx.sh ${train_output_path} inference inference_onnx hifigan_vctk
fi
# inference with onnxruntime, use fastspeech2 + hifigan by default
if [ ${stage} -le 6 ] && [ ${stop_stage} -ge 6 ]; then
./local/ort_predict.sh ${train_output_path}
fi
examples/vctk/voc1/README.md
浏览文件 @ 0fa3fdb9
......@@ -135,6 +135,13 @@ optional arguments:
Pretrained models can be downloaded here:
- [pwg_vctk_ckpt_0.1.1.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwg_vctk_ckpt_0.1.1.zip)
The static model can be downloaded here:
- [pwgan_vctk_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwgan_vctk_static_1.1.0.zip)
The ONNX model can be downloaded here:
- [pwgan_vctk_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwgan_vctk_onnx_1.1.0.zip)
Parallel WaveGAN checkpoint contains files listed below.
```text
......
examples/vctk/voc5/README.md
浏览文件 @ 0fa3fdb9
......@@ -121,6 +121,12 @@ optional arguments:
The pretrained model can be downloaded here:
- [hifigan_vctk_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_vctk_ckpt_0.2.0.zip)
The static model can be downloaded here:
- [hifigan_vctk_static_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_vctk_static_1.1.0.zip)
The ONNX model can be downloaded here:
- [hifigan_vctk_onnx_1.1.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_vctk_onnx_1.1.0.zip)
Model | Step | eval/generator_loss | eval/mel_loss| eval/feature_matching_loss
:-------------:| :------------:| :-----: | :-----: | :--------:
......
paddlespeech/t2s/exps/inference.py
浏览文件 @ 0fa3fdb9
......@@ -35,8 +35,12 @@ def parse_args():
type=str,
default='fastspeech2_csmsc',
choices=[
'speedyspeech_csmsc', 'fastspeech2_csmsc', 'fastspeech2_aishell3',
'fastspeech2_vctk', 'tacotron2_csmsc'
'speedyspeech_csmsc',
'fastspeech2_csmsc',
'fastspeech2_aishell3',
'fastspeech2_ljspeech',
'fastspeech2_vctk',
'tacotron2_csmsc',
],
help='Choose acoustic model type of tts task.')
parser.add_argument(
......@@ -56,8 +60,16 @@ def parse_args():
type=str,
default='pwgan_csmsc',
choices=[
'pwgan_csmsc', 'mb_melgan_csmsc', 'hifigan_csmsc', 'pwgan_aishell3',
'pwgan_vctk', 'wavernn_csmsc'
'pwgan_csmsc',
'pwgan_aishell3',
'pwgan_ljspeech',
'pwgan_vctk',
'mb_melgan_csmsc',
'hifigan_csmsc',
'hifigan_aishell3',
'hifigan_ljspeech',
'hifigan_vctk',
'wavernn_csmsc',
],
help='Choose vocoder type of tts task.')
# other
......
paddlespeech/t2s/exps/ort_predict_e2e.py
浏览文件 @ 0fa3fdb9
......@@ -54,19 +54,31 @@ def ort_predict(args):
device=args.device,
cpu_threads=args.cpu_threads)
merge_sentences = True
# frontend warmup
# Loading model cost 0.5+ seconds
if args.lang == 'zh':
frontend.get_input_ids("你好，欢迎使用飞桨框架进行深度学习研究！", merge_sentences=True)
frontend.get_input_ids(
"你好，欢迎使用飞桨框架进行深度学习研究！", merge_sentences=merge_sentences)
else:
print("lang should in be 'zh' here!")
frontend.get_input_ids(
"hello, thank you, thank you very much",
merge_sentences=merge_sentences)
# am warmup
spk_id = [args.spk_id]
for T in [27, 38, 54]:
am_input_feed = {}
if am_name == 'fastspeech2':
if args.lang == 'en':
phone_ids = np.random.randint(1, 78, size=(T, ))
else:
phone_ids = np.random.randint(1, 266, size=(T, ))
am_input_feed.update({'text': phone_ids})
if am_dataset in {"aishell3", "vctk"}:
am_input_feed.update({'spk_id': spk_id})
elif am_name == 'speedyspeech':
phone_ids = np.random.randint(1, 92, size=(T, ))
tone_ids = np.random.randint(1, 5, size=(T, ))
......@@ -96,12 +108,18 @@ def ort_predict(args):
phone_ids = input_ids["phone_ids"]
if get_tone_ids:
tone_ids = input_ids["tone_ids"]
elif args.lang == 'en':
input_ids = frontend.get_input_ids(
sentence, merge_sentences=merge_sentences)
phone_ids = input_ids["phone_ids"]
else:
print("lang should in be 'zh' here!")
print("lang should in {'zh', 'en'}!")
# merge_sentences=True here, so we only use the first item of phone_ids
phone_ids = phone_ids[0].numpy()
if am_name == 'fastspeech2':
am_input_feed.update({'text': phone_ids})
if am_dataset in {"aishell3", "vctk"}:
am_input_feed.update({'spk_id': spk_id})
elif am_name == 'speedyspeech':
tone_ids = tone_ids[0].numpy()
am_input_feed.update({'phones': phone_ids, 'tones': tone_ids})
......@@ -130,19 +148,40 @@ def parse_args():
'--am',
type=str,
default='fastspeech2_csmsc',
choices=['fastspeech2_csmsc', 'speedyspeech_csmsc'],
choices=[
'fastspeech2_csmsc',
'fastspeech2_aishell3',
'fastspeech2_ljspeech',
'fastspeech2_vctk',
'speedyspeech_csmsc',
],
help='Choose acoustic model type of tts task.')
parser.add_argument(
"--phones_dict", type=str, default=None, help="phone vocabulary file.")
parser.add_argument(
"--tones_dict", type=str, default=None, help="tone vocabulary file.")
parser.add_argument(
'--spk_id',
type=int,
default=0,
help='spk id for multi speaker acoustic model')
# voc
parser.add_argument(
'--voc',
type=str,
default='hifigan_csmsc',
choices=['hifigan_csmsc', 'mb_melgan_csmsc', 'pwgan_csmsc'],
choices=[
'pwgan_csmsc',
'pwgan_aishell3',
'pwgan_ljspeech',
'pwgan_vctk',
'hifigan_csmsc',
'hifigan_aishell3',
'hifigan_ljspeech',
'hifigan_vctk',
'mb_melgan_csmsc',
],
help='Choose vocoder type of tts task.')
# other
parser.add_argument(
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或