FIDA Normalizer

A text normalization package for TTS (Text-to-Speech) preprocessing with multi-language support.

Features

• Multi-language Support: Supports Italian, English, French, German, Spanish, Portuguese, Dutch, and Swedish
• Text Normalization: Converts numbers, dates, times, emails, URLs, and special characters to spoken forms
• Phonemization Support: Optional IPA phoneme generation using Espeak
• TTS Integration: Designed to work with NeMo and other TTS systems
• Currency & Units: Handles currency symbols and measurement units

Installation

From PyPI (Recommended)

pip install fida-normalizer

From Source

git clone <repository-url>
cd fida_tts
pip install -e .

For Development

pip install -e ".[dev]"

Usage

Basic Usage

from Normalizer import Normalizer

# Create a normalizer instance (Italian language by default)
normalizer = Normalizer(lang='it')

# Normalize text
result = normalizer.normalize("Il prezzo è 50,50 euro")
print(result)  # Output: il prezzo è cinquanta virgola cinquanta euro

With Different Languages

from Normalizer import Normalizer

# English normalizer
en_normalizer = Normalizer(lang='en')
result = en_normalizer.normalize("The price is $100.50")
print(result)  # Output: the price is dollar one hundred point fifty

With Phonemization

from Normalizer import Normalizer

# Enable phonemization (requires espeak/espeak-ng installed)
normalizer = Normalizer(lang='it', phonemize=True)
result = normalizer.normalize("Ciao mondo")
print(result)  # Output: IPA phonemes

Configuration Options

from Normalizer import Normalizer

normalizer = Normalizer(
    lang='it',           # Language code: 'it', 'en', 'fr', 'de', 'es', 'pt', 'nl', 'sv'
    tts_mode=True,       # TTS mode: keeps punctuation suitable for TTS
    to_lower=True,       # Convert output to lowercase
    phonemize=False      # Enable/disable phonemization
)

Supported Languages

Supported Transformations

• Numbers: Converts digits to spoken words (e.g., "123" → "one hundred twenty-three")
• Decimals: Handles decimal separators based on locale
• Percentages: Converts "%" to spoken form
• Currency: Handles $, €, £, ¥ symbols
• Dates: Converts date formats to spoken form
• Times: Converts time formats (HH:MM, HH:MM:SS)
• Emails: Spells out email addresses
• URLs/Domains: Spells out web addresses
• Units: Converts measurement units (m, kg, km/h, etc.)
• Special Characters: Handles @, &, -, _, /, etc.

Requirements

• Python >= 3.8
• phonemizer >= 3.0.0 (optional, for phonemization)
• espeak or espeak-ng (system dependency for phonemization)

License

MIT License

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.