Speakr is a personal, self-hosted web application designed for transcribing audio recordings

Udacity NLP nano degree projects

Exploring Biases in Transformer-Based End-to-End ASR Systems

An implementation of a DNN speech recognizer as part of the Udacity NLP NanoDegree program

Ed-Fed: A generic federated learning framework for edge devices

Implement a deep neural network that functions as part of an end-to-end automatic speech recognition (ASR) pipeline

基于 FastAPI、Vue3、Whisper 和 DeepSeek API 的 AI 语音对联互动系统，支持语音输入、自动生成下联和智能评分。An AI-powered interactive Chinese couplet system based on FastAPI, Vue3, Whisper, and DeepSeek API. Supports voice input, automatic couplet generation, and intelligent evaluation.

IF4072 Natural Language & Text Processing

Comprehensive benchmark of OpenAI Whisper models for Bosnian, Croatian, and Serbian languages. Includes pipelines for audio transcription, rigorous text normalization, Levenshtein distance evaluation, and LLM-based post-processing.

A web game based on Taiwanese tourism. Using Python as backend.

WhisperX ASR is a FastAPI-based application for automatic speech recognition. It transcribes audio files to text using WhisperX, supports multiple languages, batch processing, and offers both a web UI and REST API.

🎤 Enable real-time speech recognition with WhisperX using FastAPI for efficient, scalable audio processing.

Thesis Project

A lightweight automatic speech recognition (ASR) and speech translation web application

tfjs

Automatic Speech Recognition (ASR) Assisted Online Text Editor

FastAPI server for locally serving Gemma 3 270M & OpenAI Whisper with batched inference and streaming support.

The dataset comprises of 5000 hours speech corpus in Akan, Ewe, Dagbani, Daagare, and Ikposo. Each language includes 1000 hours of audio speech from indigenous speakers of the language. Of which 100 hours is transcribed.