Logo Lanfrica
  • Accueil
  • Atlas
  • Analyses
  • Documentation
  • Sign in

© 2026 Lanfrica. Tous droits réservés. Tous les droits d'auteur des ressources affichées sur le site Web Lanfrica appartiennent aux détenteurs de droits d'auteur d'origine, sauf indication contraire explicite.

frank-shema/Kinya-whisper-work

Domaine:

natural language processing

Type de record:

software
Créateur:
fra
Hôte:
A voice-enabled Kinyarwanda assistant that uses a fine-tuned Whisper model for automatic speech recognition (ASR), rule-based NLP with fuzzy matching, and offline text-to-speech (TTS). Built for the Intelligent Robotics course at Rwanda Coding Academy, it demonstrates real-time and batch voice interaction in Kinyarwanda. 🗣️ KinyaWhisper – Kinyarwanda Voice Assistant A lightweight, voice-enabled assistant that understands and responds in Kinyarwanda, simulating humanoid robot voice interaction. Built for the Intelligent Robotics course at Rwanda Coding Academy, this project leverages Automatic Speech Recognition (ASR), Natural Language Processing (NLP), and Text-to-Speech (TTS) to create an AI-powered assistant tailored for Kinyarwanda speakers. 🎯 Project Overview KinyaWhisper is designed to: 🎤 Transcribe Kinyarwanda speech using a fine-tuned OpenAI Whisper model. 🧠 Understand questions via rule-based NLP with fuzzy matching. 🗣️ Respond in Kinyarwanda using offline TTS (pyttsx3). This project demonstrates the potential of localized AI for voice interaction in underrepresented languages like Kinyarwanda. 📁 Project Structure . ├── audio/ # Custom Kinyarwanda audio samples (44 WAV files) ├── dataset.jsonl # Metadata for training (audio paths + transcriptions) ├── kinya-whisper-model/ # Fine-tuned Whisper model ├── train.py # Script for fine-tuning Whisper ├── inference.py # Script for batch transcription ├── main.py # Batch-mode voice assistant (ASR + NLP + TTS) ├── bach_main.py # CLI for live audio recording and response ├── transcriptions.txt # Output transcriptions from inference ├── README.md # Project documentation 🚀 Features ASR: Converts Kinyarwanda speech to text using a fine-tuned Whisper model. NLP: Matches transcribed text to predefined answers using fuzzy logic (difflib). TTS: Generates spoken responses in Kinyarwanda via pyttsx3. Live Recording: Supports real-time voice input via microphone (bach_main.py). Batch Processing: Processes multiple audio files for transcription and response (main.py). 🛠️ Technologies Python 3.10+ OpenAI Whisper (fine-tuned whisper-small) Transformers (Hugging Face) Torchaudio for audio processing Pyttsx3 for offline TTS Sounddevice for live microphone input Difflib for fuzzy text matching 🧠 Model Training Dataset: 44 custom Kinyarwanda …

Visit

github.com

Tasks

automatic speech recognitionspeech processingtext to speech

Languages

Kinyarwanda

Tags

python

Similaires

SHEMA-YVAN/menya-2.0Rutabana/kinyarwadataforum/kinya-performanceijuru/kinya-siteheritierbagumire/Kinya-AI-assistantSIGYE/Kinya-voice-assistant

SHEMA-YVAN/menya-2.0

# MENYA — Learn. Practice. Excel. A centralized digital learning platform for Rwandan learners from

Rutabana/kinya

Kinyarwanda learning app — flashcards, translation, daily radio This is a Next.js project bootstrap

rwadataforum/kinya-performance

a repo of ideas to test the performance of AI models in Kinyarwanda # Kinyarwanda Performance Evalu

ijuru/kinya-site

Work in progress for next version of kinyarwanda.net Kinyarwanda.net Site Source ==================

heritierbagumire/Kinya-AI-assistant

his project shows how intelligent robots can communicate in local languages, empowering them to assi

SIGYE/Kinya-voice-assistant

A Kinyarwanda voice assistant # 🗣️ Kinyarwanda Voice Assistant A bilingual voice interaction tool