International News Reader with translation - over 200 international news feeds included

An open-source, educational app for speech-to-text & text translation that runs entirely in your browser: record or upload audio, transcribe with Whisper-style models, translate into many languages, & export results—without sending audio or transcripts to an app server for processing. Heavy ML runs in Web Workers via Transformers.js & ONNX Runtime

AfriNLLB: Efficient Translation Models for African Languages

An automated, end-to-end AI video dubbing pipeline. Translates dialogue, clones speaker voices zero-shot, isolates background SFX, and mathematically syncs audio to create studio-quality cinematic dubs. Built with Whisper, Pyannote, XTTS, and Demucs.

A high-performance local AI pipeline for restoring VHS audio, transcribing with Whisper, and translating subtitles using NLLB-200.

THOTH Language Translator is a powerful, privacy-first, state-of-the art translation tool that runs offline. It uses the NLLB-200 engine to translate over 200 languages, with option to switch to an alternate translation engine, Argos.

A two-stage QLoRA fine-tuning framework for high-fidelity Twi-to-English translation using Meta's NLLB-200. Explores large-scale synthetic pre-training and human-aligned refinement paradigms under strict 6GB VRAM hardware constraints

Home Lab POC - Local offline capable AI pipeline for voice transcription, multilingual translation and summarization

Plateforme AgriTech intégrée pour les agriculteurs sénégalais — Stockage connecté · Warrantage numérique · Mise en marché · Chatbot IA · 4 langues (FR/EN/AR/Wolof)

A privacy-first, fully offline neural machine translation application supporting 200+ languages. Built on Meta's NLLB-200 model, this tool runs entirely in your browser without requiring server infrastructure or internet connectivity after initial model download.

LoRA fine-tuning of NLLB-200 (1.3B) for Chinese-to-English translation on WMT19, reaching 24.9 BLEU.

A real-time audio streaming POC featuring Voice Activity Detection (VAD), Faster-Whisper ASR, NLLB-200 translation, and Piper TTS. Built with FastAPI and React to demonstrate a low-latency, end-to-end speech-to-speech pipeline.

Comparing mBART-50, NLLB-200, Helsinki-NLP & mT5 for English→Malayalam NMT using Back-Translation data augmentation in low-resource settings.

NLLB-200 modelini kullanarak Türkçe-İngilizce ve İspanyolca dilleri arasında metin çevirisi yapan basit bir Streamlit uygulaması. A simple Streamlit app for text translation between Turkish, English and Spanish languages using the NLLB-200 model.

Offline multi-language receipt scanner and split bill engine using PaddleOCR and NLLB-200

Fine-tuning NLLB-200 for Hinglish & Spanglish → English translation using the LinCE benchmark — with training, BLEU/ChrF/COMET evaluation, batch inference, and ONNX export.

A machine translation evaluation pipeline prioritizing Meta's NLLB-200 model and the FLORES-200 dataset, featuring Kaggle-optimized data loading and UMAP embedding visualization.

CourtAccess AI is an MLOps platform providing real-time speech translation and legal document processing for the Massachusetts Trial Court, ensuring equitable access to justice for Limited English Proficient individuals.

WeWhisper-AI is a 100% free desktop app offering transcription, speaker diarization, and translation powered by open-source models. Available on Windows, Linux, and macOS with CUDA and Metal acceleration. Ideal for privacy-conscious users, it enables transcription, dictation, and translation of audio files and recordings directly on your machine.