indicTrans
indicTranslate v1 - Machine Translation for 11 Indic languages. For latest v2, check: https://github.com/AI4Bharat/IndicTrans2
- 145
- Jupyter Notebook
- MIT
- 3 y ago
indicTranslate v1 - Machine Translation for 11 Indic languages. For latest v2, check: https://github.com/AI4Bharat/IndicTrans2
A configurable engine for analysing multi-lingual and multi-modal content.
Open, self-hosted voice cloning for Indian languages that keeps a speaker's accent and dialect. Built for archival recordings, with consent and an inaudible watermark. प्रतिध्वनि: Sanskrit for "echo".
This repositary hosts my experiments for the project, I did with OffNote Labs.
MMT: A Multilingual and Multi-Topic Indian Social Media Dataset (C3NLP / EACL 2023). Samples for multilingual topic modeling, language identification, code-mixing and Hinglish.
This repository demonstrates usage of Amazon Bedrock Claude 3 models for Indian languages. The use cases include but not limited to: 1. Information extraction, 2. Question answering 3. Summarisation, 4. Translation and 5. Transliteration from the content in Indian languages such as Hindi, Telugu, Tamil, Malayalam, Marathi, Kannada to mention a few.
Self-hosted real-time voice agent platform: Pipecat runtime, local Whisper and Kokoro, provider switching, per-conversation latency and cost, privacy by default, Indian languages.
Train a Llama-style Hindi LLM from scratch on a single 6 GB GPU: SentencePiece tokenizer, packed data pipeline, plain PyTorch training. Extensible to 10+ Indian languages.
Cross-Lingual Gaps in Visual STEM Reasoning (MRL 2026, ACL Anthology): 8 VLMs audited across 6 Indian languages on MathVista, ScienceQA and MMMU
OmniVoice TTS server that speaks math, LaTeX, ₹ amounts and educational content correctly. Voice cloning, 600+ languages, Hindi-ready.