Cantonese/Chinese Text to Speech based on statistical parametric speech synthesis using merlin toolkit

This project is influenced by MTTS

1. First, you need data contain wav and txt (prosody mark is optional)
2. Second, generate HTS label using this project
3. Using merlin/egs/cantonese_voice to train and generate Cantonese Voice

• Context related annotation
• Question Set
• Rules to design a Question Set

Python : python3.6
System: linux(tested on ubuntu16.04)

sudo apt-get install libatlas3-base

Run bash tools/install_mtts.sh
Or download file by yourself

• Download montreal-forced-aligner and unzip to directory tools/

Run Demo

bash run_demo.sh

• Usage: Run python src/mtts.py txtfile wav_directory_path output_directory_path (Absolute path or relative path) Then you will get HTS label, if you have your own acoustic model trained by monthreal-forced-aligner, add-a your_acoustic_model.zip, otherwise, this project use thchs30.zip acoustic model as default
• Attention: Currently only support Chinese Character, txt should not have any Arabia number or English alphabet

txtfile example

A_01 这是一段文本
A_02 这是第二段文本

wav_directory example(Sampleing Rate should larger than 16khz)

A_01.wav  
A_02.wav  

• Usage: Run python src/mandarin_frontend.py txtfile output_directory_path
• or import mandarin_frontend

from mandarin_frontend import txt2label

result = txt2label('向香港特别行政区同胞澳门和台湾同胞海外侨胞')
[print(line) for line in result]

see source code for more information, but pay attention to the alignment file(sfs file), the format is endtime phone_type not start_time, phone_type(which is different from speech ocean's data)

This project use Montreal-Forced-Aligner to do forced alignment, if you want to get a better alignment, use your data to train a alignment-model, see mfa: algin-using-only-the-dataset

1. We trained the acoustic model on our dataset.

You can generate HTS Label without prosody mark. we assume that word segment is smaller than prosodic word(which is adjusted in code)

• Text Normalization
• Better Chinese word segment
• G2P: Polyphone Problem
• Better Label format and Question Set
• Improvement of prosody analyse
• Better alignment

• miran899