g2pM - Neural Grapheme-to-Phoneme for Mandarin Chinese

0.1.2.5 verified Sat May 09 auth: no python maintenance

g2pM is a neural grapheme-to-phoneme conversion package for Mandarin Chinese, leveraging a transformer-based model to convert Chinese characters (pinyin) to phonetic symbols (bopomofo/zhuyin). Current version 0.1.2.5, released infrequently; last update in 2021.

pip install g2pm

Common errors

error ModuleNotFoundError: No module named 'g2pm' ↓

cause Package not installed or misspelled; sometimes confused with 'g2p' or 'g2p-en'.

fix

Run: pip install g2pm

error AttributeError: module 'g2pm' has no attribute 'G2pM' ↓

cause Incorrect import pattern (e.g., 'import g2pm' then 'g2pm.G2pM()' but G2pM is not a direct attribute of the module).

fix

Use: from g2pm import G2pM

error FileNotFoundError: [Errno 2] No such file or directory: '.../g2pm/model/model.pb' ↓

cause Model file not downloaded or corrupted because download interrupted.

fix

Delete the cached model directory (usually ~/.g2pm/) and re-initialize: from g2pm import G2pM; g2p = G2pM()

Warnings

gotcha Input must be pinyin with tone numbers (e.g., 'ni3'), not Chinese characters or pinyin without tones. The model does not handle characters. ↓

fix Preprocess your text to extract pinyin syllables with tone numbers before using g2pM.

gotcha The package downloads a pre-trained model on first use (~75 MB). Ensure network connectivity or manually cache the model. ↓

fix Run g2p = G2pM() once to trigger download; subsequent calls use cached model.

deprecated The package has not been updated since 2021 and may not work with newer Python versions (e.g., 3.10+). Known issues with tensorflow or torch dependency mismatches. ↓

fix Consider using an alternative like pypinyin or jieba for Chinese text processing.

Imports

G2pM
wrong
```
import g2pm
```
correct
```
from g2pm import G2pM
```
G2pM is a class in the g2pm module, not the module itself.

Quickstart

Initialize G2pM and convert pinyin to bopomofo.

from g2pm import G2pM

g2p = G2pM()
# Convert pinyin with tone numbers to bopomofo
result = g2p('wo3')
print(result)  # Output: ㄨㄛˇ

# For a list of pinyin strings
results = g2p(['ni3', 'hao3'])
print(results)  # Output: ['ㄋㄧˇ', 'ㄏㄠˇ']