Language Technology

Collaborative catalog of resources for Indic language NLP.

Indic computing resources with a focus on Telugu.

Normalization, tokenization and script utilities for Indian languages.

Stanford NLP pipeline with Telugu models.

Finite-state script normalization and processing from Google Research.

Part-of-speech tagger and chunker for Indian languages.

Named entity recognition from IIIT Hyderabad's LTRC.

Language identification for native and romanized Indian text.

Tokenizer for Telugu.

A 278M Telugu language model trained from scratch.

Gemma fine-tuned for Indic languages by Telugu LLM Labs.

Open language model optimized for Indian languages.

Multilingual BERT for Indian languages.

Community corpus-collection project from Swecha.

Universal Dependencies treebank.

Code-switched Universal Dependencies treebank.

Native and romanized text for South Asian languages.

Large transliteration dataset.

Bharat Parallel Corpus Collection for translation.

Instruction and safety datasets for Indic LLMs.

Question answering.

Sentiment classification.

Named entity recognition.

Headline generation.

Paraphrase generation.

Instruction data in native and romanized Telugu.

Classical Telugu texts packaged for the Classical Language Toolkit.

Scraper for building a news corpus.

Scraper for text from Telugu novels.

Natural speech dataset across 22 Indian languages.

Speech corpus for Indian languages.

Crowdsourced multi-speaker Telugu speech dataset.

Common words spoken by native speakers from Andhra Pradesh and Telangana.

Speech recognition for Indian languages.

Pretraining and fine-tuning scripts for Indic speech recognition.

Multilingual Indic speech recognition with per-language LoRA adapters.

Telugu speech recognition model.

On-device streaming speech-to-text.

Text-to-speech for languages of India.

Multilingual text-to-speech model.

Text-to-speech controlled with natural language descriptions.

VITS voice from Meta's Massively Multilingual Speech project, under a non-commercial license.

Voices for the Piper local text-to-speech engine.

Trained LSTM model for Tesseract.

Recent OCR model for Telugu.

End-to-end OCR based on convolutional networks.

Character recognition using CNNs.

Python library for metrical poetry.

Machine verification and identification of metrical poetry.

Detects and classifies classical poetic meters.