Revo Norm¶

Text normalization for Text-to-Speech, built for English and Malay (Bahasa Melayu).

Revo Norm converts written text — numbers, currency, dates, times, abbreviations, acronyms, and more — into natural spoken form so your TTS engine reads it the way a human would say it.

30-Second Quickstart¶

from revo_norm import normalize_text

# English
print(normalize_text("The price is RM50K due on 15/08/2025", language="en"))
# "The price is fifty thousand ringgit due on fifteen of August twenty twenty-five"

# Malay
print(normalize_text("Suhu 25C hari ini", language="ms"))
# "Suhu dua puluh lima celcius hari ini"

# Minimal — just fix spacing
print(normalize_text("Hello   world", language="en", profile="minimal"))
# "Hello world"

Key Features¶

Currency normalization — RM50K, $1.5M, USD 200B expanded and spoken
Date recognition — 15/08/2025, 2025-08-15, 15 August 2025
Time recognition — 3:30 pm, 14:00
Number-to-words — cardinal and ordinal numbers in English and Malay
Acronym handling — IBM → "I B M", NASA preserved, JSON → "J son"
Temperature — 25C, 100°F converted to spoken form
Fractions — 3/4 → "three over four"
Measurements — 5km, 10kg, 3GB
Entity protection — currency, URLs, emails, and dates shielded from cascading transforms
Pronunciation mappings — custom overrides with highest pipeline priority (GUI → "gooey")
Bilingual — full support for English (en) and Malay (ms)
Configurable profiles — minimal, basic, standard, aggressive presets

Installation¶

# pip
pip install revo-norm

# uv (recommended)
uv add revo-norm

See the Installation Guide for all options including source installs and dev setup.

Documentation¶

Page	Description
Installation	Install via pip, uv, or from source
Quickstart Guide	5-minute walkthrough with examples
API Reference	Full API documentation
Features	Detailed feature guides

License¶

MIT