BelNorm 0.3.0

Belarusian text normalizer

Belarusian Text Normalization Library

This library provides tools for preparing Belarusian text for Text-to-Speech (TTS) applications. Key features include:

• Splitting text into sentences
• Normalizing numbers and abbreviations
• Handling common Belarusian contractions and special symbols

Designed for easy integration, this library helps ensure that text is correctly formatted and pronounced by TTS systems.

Usage example

from BelNorm import BelSplitter, NormalizerLLM
from BelG2P import BelG2PWrapper

g2p=BelG2PWrapper()
normalizer=NormalizerLLM("gemini/gemini-flash-lite-latest")
splitter=BelSplitter()

text="Нейкі беларускі тэкст. І прачытаць яго трэба 3 разы."
print(f"Input text: {text}")

for paragraph in splitter.parse([text]):
    for sentence in paragraph:
        if not sentence.is_normalized:
            print("Normalization required: "+sentence.text)
            sentence.normalize(normalizer.normalize)
            print(sentence.text)
        print(sentence.convert_tts(g2p.convert))

Do not forget to add some environment variable for LLM usage. For Gemini, it should be GEMINI_API_KEY.