Low Resource Languages Python Packages

africa-g2p

Text and orthography toolkit for 869 African languages: grapheme-to-phoneme (native spelling or IPA), conversion between any two languages' orthographies, one shared cross-language spelling, and a reversible plain-ASCII form. Non-Latin scripts romanised. Pure Python, no dependencies.

amharic-tokenizer

Syllable-aware BPE tokenizer for the Amharic language (አማርኛ) – fast, accurate, trainable.

pyidaungsu-numpy2

Tiny language detector + tokenizer for 50+ SE/South Asian languages (Burmese, Karen, Chin, Shan, Mon, Khmer, Lao, Thai, Tamil, Hindi, …). Revamp of pyidaungsu.

chandassu

Chandassu: First Python Library for Global Metrical Poetry

everyvoice

The EveryVoice TTS Toolkit - Text To Speech for your language

pyidaungsu

Tiny language detector + tokenizer for 50+ SE/South Asian languages (Burmese, Karen, Chin, Shan, Mon, Khmer, Lao, Thai, Tamil, Hindi, …). Revamp of pyidaungsu.

yotext

A Python Library for orthographic normalization and diacritic restoration for Yorùbá text. Separates tone marks from underdots. It has zero dependencies with a CLI and a web demo.

calamancy

NLP pipelines for Tagalog using spaCy

africa-speech-synth

Build synthetic speech datasets for African languages with Google Gemini TTS — corpus text, phoneme-coverage sentence selection, africa-g2p orthography normalisation, and a training-ready HuggingFace dataset

afrispeech-synth

Build synthetic speech datasets for African languages with Google Gemini TTS — corpus text, phoneme-coverage sentence selection, africa-g2p orthography normalisation, and a training-ready HuggingFace dataset

sakhaspell

Спелчекер якутского языка: 300k словоформ, восстановление ҕ ҥ ө һ ү, 1.33% ложных срабатываний

african-speech-mcp

Quality-aware, read-only MCP server for discovering and selecting African speech corpora, Wolof first, with audited metrics, provenance, licensing, and training-set planning.

windytranslate

Python client for Windstorm Labs' open translation and speech models: find, license-check and run them locally. windytranslate.com

tojiki

Open-source tools for the Tajik language: Cyrillic↔Latin transliteration, numbers to words, text normalization

glossobench

Open-source, judge-agnostic LLM benchmark for any language — natural or AI-developed. Turn hours to minutes, save days of benchmark cost! Public data, no gate, $0 offline evaluation.

ricelang

Tiny language detector + tokenizer for 50+ SE/South Asian languages (Burmese, Karen, Chin, Shan, Mon, Khmer, Lao, Thai, Tamil, Hindi, …). Revamp of pyidaungsu.

lingualdub

An open, modular framework for building and evaluating low-resource speech-AI pipelines

padie-extended

An open-source Python package that uses AI to predict Nigerian languages, including English, Pidgin, Yoruba, Hausa, and Igbo.

meitei-senter

Neural sentence boundary detection for Meitei Mayek (Manipuri) using SentencePiece tokenization and a CNN-based spaCy pipeline.

darija-tools

Dependency-free Python utilities for Darija normalization and bidirectional Arabizi transliteration.

annaparavai

This repository contains the implementation of a transfer learning-based approach for detecting AI-generated product reviews in Tamil and Malayalam. It includes pretrained model embeddings, deep neural networks, and an ensemble method to enhance classification accuracy.

ssl-kaldi-feats

Kaldi recipes that feed frozen self-supervised encoder features (HuBERT, WavLM, MMS, mHuBERT, AV-HuBERT, etc.) to GMM-HMM, TDNN-F chain and flat-start e2e chain models. Multilingual ASR recipes, 3.88% WER on Mini LibriSpeech, and GRID lipreading at 5.85% WER on unseen speakers.

ne-agent

Terminal AI agent for Northeast India's low-resource languages. Local, offline, multilingual routing (search, translate, transcribe).

ne-embed

Multilingual sentence embeddings for 10 Northeast Indian languages