English | 简体中文

PaddleSpeech

Quick Start | Tutorials | Models List

---

PaddleSpeech is an open-source toolkit on PaddlePaddle platform for two critical tasks in Speech - Automatic Speech Recognition (ASR) and Text-To-Speech Synthesis (TTS), with modules involving state-of-art and influential models.

Via the easy-to-use, efficient, flexible and scalable implementation, our vision is to empower both industrial application and academic research, including training, inference & testing module, and deployment. Besides, this toolkit also features at:

• Fast and Light-weight: we provide a high-speed and ultra-lightweight model that is convenient for industrial deployment.
• Rule-based Chinese frontend: our frontend contains Text Normalization (TN) and Grapheme-to-Phoneme (G2P, including Polyphone and Tone Sandhi). Moreover, we use self-defined linguistic rules to adapt Chinese context.
• Varieties of Functions that Vitalize Research:
  • Integration of mainstream models and datasets: the toolkit implements modules that participate in the whole pipeline of both ASR and TTS, and uses datasets like LibriSpeech, LJSpeech, AIShell, etc. See also model lists for more details.
  • Support of ASR streaming and non-streaming data: This toolkit contains non-streaming/streaming models like DeepSpeech2, Transformer, Conformer and U2.

Let's install PaddleSpeech with only a few lines of code!

Note: The official name is still deepspeech. 2021/10/26

# 1. Install essential libraries and paddlepaddle first.
# install prerequisites
sudo apt-get install -y sox pkg-config libflac-dev libogg-dev libvorbis-dev libboost-dev swig python3-dev libsndfile1
# `pip install paddlepaddle-gpu` instead if you are using GPU. 
pip install paddlepaddle

# 2.Then install PaddleSpeech.
git clone https://github.com/PaddlePaddle/DeepSpeech.git
cd DeepSpeech
pip install -e .

Table of Contents

The contents of this README is as follow:

• Alternative Installation
• Quick Start
• Models List
• Tutorials
• FAQ and Contributing
• License
• Acknowledgement

Alternative Installation

The base environment in this page is

• Ubuntu 16.04
• python>=3.7
• paddlepaddle==2.1.2

If you want to set up PaddleSpeech in other environment, please see the ASR installation and TTS installation documents for all the alternatives.

Quick Start

Note: ckptfile should be replaced by real path that represents files or folders later. Similarly, exp/default is the folder that contains the pretrained models.

Try a tiny ASR DeepSpeech2 model training on toy set of LibriSpeech:

cd examples/tiny/s0/
# source the environment
source path.sh
# prepare librispeech dataset
bash local/data.sh
# evaluate your ckptfile model file
bash local/test.sh conf/deepspeech2.yaml ckptfile offline

For TTS, try FastSpeech2 on LJSpeech:

• Download LJSpeech-1.1 from the ljspeech official website and our prepared durations for fastspeech2 ljspeech_alignment.
• Assume your path to the dataset is ~/datasets/LJSpeech-1.1 and ./ljspeech_alignment accordingly, preprocess your data and then use our pretrained model to synthesize:

bash ./local/preprocess.sh conf/default.yaml
bash ./local/synthesize_e2e.sh conf/default.yaml exp/default ckptfile

If you want to try more functions like training and tuning, please see ASR getting started and TTS Basic Use.

Models List

PaddleSpeech ASR supports a lot of mainstream models, which are summarized as follow. For more information, please refer to ASR Models.

ASR Module Type	Dataset	Model Type	Link
Acoustic Model	Aishell	2 Conv + 5 LSTM layers with only forward direction	Ds2 Online Aishell Model
		2 Conv + 3 bidirectional GRU layers	Ds2 Offline Aishell Model
		Encoder:Conformer, Decoder:Transformer, Decoding method: Attention + CTC	Conformer Offline Aishell Model
		Encoder:Conformer, Decoder:Transformer, Decoding method: Attention	Conformer Librispeech Model
	Librispeech	Encoder:Conformer, Decoder:Transformer, Decoding method: Attention	Conformer Librispeech Model
		Encoder:Transformer, Decoder:Transformer, Decoding method: Attention	Transformer Librispeech Model
Language Model	CommonCrawl(en.00)	English Language Model	English Language Model
	Baidu Internal Corpus	Mandarin Language Model Small	Mandarin Language Model Small
		Mandarin Language Model Large	Mandarin Language Model Large

PaddleSpeech TTS mainly contains three modules: Text Frontend, Acoustic Model and Vocoder. Acoustic Model and Vocoder models are listed as follow:

TTS Module Type	Model Type	Dataset	Link
Text Frontend			chinese-fronted
Acoustic Model	Tacotron2	LJSpeech	tacotron2-vctk
	TransformerTTS		transformer-ljspeech
	SpeedySpeech	CSMSC	speedyspeech-csmsc
	FastSpeech2	AISHELL-3	fastspeech2-aishell3
		VCTK	fastspeech2-vctk
		LJSpeech	fastspeech2-ljspeech
		CSMSC	fastspeech2-csmsc
Vocoder	WaveFlow	LJSpeech	waveflow-ljspeech
	Parallel WaveGAN	LJSpeech	PWGAN-ljspeech
		VCTK	PWGAN-vctk
		CSMSC	PWGAN-csmsc
Voice Cloning	GE2E	AISHELL-3, etc.	ge2e
	GE2E + Tactron2	AISHELL-3	ge2e-tactron2-aishell3

Tutorials

Normally, Speech SoTA gives you an overview of the hot academic topics in speech. If you want to focus on the two tasks in PaddleSpeech, you will find the following guidelines are helpful to grasp the core ideas.

The original ASR module is based on Baidu's DeepSpeech which is an independent product named DeepSpeech. However, the toolkit aligns almost all the SoTA modules in the pipeline. Specifically, these modules are

• Data Prepration
• Data Augmentation
• Ngram LM
• Benchmark
• Relased Model

The TTS module is originally called Parakeet, and now merged with DeepSpeech. If you are interested in academic research about this function, please see TTS research overview. Also, this document is a good guideline for the pipeline components.

FAQ and Contributing

You are warmly welcome to submit questions in discussions and bug reports in issues! Also, we highly appreciate if you would like to contribute to this project!

License

PaddleSpeech is provided under the Apache-2.0 License.

Acknowledgement

PaddleSpeech depends on a lot of open source repos. See references for more information.