ACP: Automatic Reconstruction of Ancient Chinese Pronunciations

This repository contains the source code for the paper "Automatic Reconstruction of Ancient Chinese Pronunciations".

Overview

This project aims to reconstruct ancient Chinese pronunciations using a transformer-based model, incorporating both glyph and temporal features. The dataset, ACP (Ancient Chinese Phonology), consists of digitized phonology rules for a large number of Chinese characters spanning multiple historical periods.

We have also created a demo website where you can see the predicted future pronunciations of some Chinese characters using our model. Check it out at Ancient Chinese Demo.

Requirements

To run the code, you'll need the following dependencies:

• python: 3.7.6
• pytorch: 1.5.0
• transformers: 3.5.0

You can install the required packages using:

pip install -r requirements.txt

Dataset

The dataset used in this project is the Ancient Chinese Phonology (ACP) dataset, which includes phonological data for a substantial number of Chinese characters.

Directory Structure

The repository is structured as follows:

├── dataset/                # Directory containing the dataset
├── train/               # Directory containing training scripts
├── utils/               # Utility scripts and helper functions
├── requirements.txt     # Required Python packages
├── README.md            # This README file
└── LICENSE              # License file

Training

To train the model, follow these steps:

1. Navigate to the train directory:

   cd train

2. Run the training script:

   python train.py

Demo Results

Below are some examples of the predicted pronunciations for various Chinese characters across different historical periods:

Character	MiddleTang	LateTang	Song	Yuan	MingQing	Modern	AD2300
觊	k - i -	k - i -	k - i -	k - əi -	kʰ - i -	ʨ - i -	ʨʰ - əi -
涶	tʰ u ɑ -	tʰ u ɑ -	tʰ u ɔ -	tʰ u ɔ -	tʰ u o -	tʰ u o -	tʰ u a -
惭	ʣ - ɑ m	z - a m	ʦʰ - a m	ʦʰ - a m	ʦʰ - a n	ʦʰ - a n	tʰ - a n
亟	kʰ - i -	kʰ - i -	kʰ - i -	tʰ - i -	t - i -	ʨ - i -	ʨʰ - i -
调	d i æu -	d i æu -	tʰ i æu -	tʰ i æu -	tʰ i ɑu -	tʰ i ɑu -	tʰ i ɔu -
禾	ɣ u ɑ -	ɦ u ɑ -	h u ɔ -	h ɛ ɔ -	x ɛ e -	x - ɤ -	x - i -
眈	d - ɑ m	d - a m	tʰ - a m	t - a n	t - a ɲ	t - a n	t - a ŋ
颐	j - i -	j - i -	j - i -	j - ɛ -	j - i -	j - i -	j - i -
淑	ʑ i o k	ʑ - u k	ɕ - u k	ɕ - o -	ʂ - o -	ʂ - u -	ʃ - u -
卧	ŋ u ɑ -	ŋ u ɑ -	ŋ u ɔ -	w - ɔ -	w u ɔ -	w - o -	w - ɔ -
畔	b u ɑ n	b u a n	p u a n	pʰ u a n	pʰ - a n	pʰ - a n	pʰ - a n
穹	kʰ i o ŋ	kʰ - u ŋ	kʰ - u ŋ	ʨʰ - u ŋ	ʂ - y ŋ	ʨʰ - y ŋ	ɣ - ua ŋ
醉	ʦ i ui -	ʦ i ui -	ʦ - ɿ -	ʦ - uə -	ʦ - uæ -	ʦ u ei -	ʦ ɚ ei -
曾	ʣ - ə ŋ	z - ə ŋ	ʦʰ - ə ŋ	ʦ - e ŋ	ʦ - i ŋ	ʦ - ə ŋ	ʦ - u ŋ
划	ɣ u ɐ k	ɦ u ə k	h u ə k	h u a k	h u ɑ -	x u ɑ -	x u a -
殂	ʣ - u -	z - u -	ʦʰ - u -	ʦ - u -	ʦ - u -	ʦʰ - u -	ʧ - u -
菱	l i ə ŋ	l i ə ŋ	l i ə ŋ	l - i ŋ	l - i ŋ	l - i ŋ	l - a ŋ
岚	l - ɑ m	l - a m	l - a m	l - a n	l - a n	l - a n	l - a ŋ
攻	k - u ŋ	k - u ŋ	k - u ŋ	k - u ŋ	k - u ŋ	k - u ŋ	kʰ - u ŋ
盲	m - ɐ ŋ	m - ɐ ŋ	m - ɐ ŋ	m - a ŋ	m - a ŋ	m - a ŋ	m - æ ŋ
觉	k - ɔ k	k - ɔ k	k - ea k	k - ea k	ʨ - ɛ -	ʨ y ɛ -	ʨ u ɛ -
莺	ʔ - ɐ ŋ	ʔ - ɐ ŋ	j - ɐ ŋ	j - ɐ ŋ	j - i ŋ	j - i ŋ	j - i ə
诂	k - u -	k - u -	k - u -	k - u -	k - u -	k - u -	kʰ - u -
荏	ȵ - i m	r i ə m	r - i m	r - ə m	r - ə n	ɻ - ə n	ɻ - æ n
核	ɣ - ɐ k	ɦ - ə k	h - ə k	f - ə k	pʰ - ə k	x - ɤ -	s - ɔ -
郝	ɕ i ɐ k	ɕ i ə k	ɕ - i t	ɕ - i tʰ	x - au -	x - ɑu -	x - ɑ -
射	j i a -	j i a -	j i a -	j i a -	ʂ - ɛ -	ʂ - ɤ -	ʂ - ə -
刊	kʰ - ɑ n	kʰ - a n	kʰ - a n	kʰ - a n	kʰ - a n	kʰ - a n	kʰ - æ n
惹	ȵ i a -	r - a -	r - a -	ɽ - e -	ɽ - e -	ɻ - ɤ -	ɻ - e -
厕	ʧʰ - i -	ʧʰ - i -	ʦʰ - ɿ -	ʦ - e -	ʦ - ɤ -	ʦʰ - ɤ -	ʦ - e -