metadata

title: Wakanda Kinyarwanda ASR
emoji: 🎤
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: 5.38.2
app_file: app.py
pinned: false
license: apache-2.0
tags:
  - speech-recognition
  - kinyarwanda
  - whisper
  - wakanda-ai
  - audio-to-text
models:
  - WakandaAI/wakanda-whisper-small-rw-v1
languages:
  - rw

🎤 Wakanda Whisper - Kinyarwanda ASR

A state-of-the-art automatic speech recognition system specifically fine-tuned for Kinyarwanda language, built on OpenAI's Whisper architecture.

🌟 Features

High Accuracy: Fine-tuned specifically for Kinyarwanda speech patterns
Multiple Input Methods: Upload audio files or record directly through microphone
Format Support: Supports WAV, MP3, M4A, FLAC, and other common audio formats
Real-time Processing: Fast inference with optimized performance
User-friendly Interface: Beautiful and intuitive web interface

🚀 Model Details

Base Architecture: OpenAI Whisper Small
Language: Kinyarwanda (rw)
Parameters: ~39M
Training Data: Curated Kinyarwanda speech dataset
Model Repository: WakandaAI/wakanda-whisper-small-rw-v1

🎯 How to Use

Option 1: Upload Audio File

Click on the "Upload Audio File" tab
Select your Kinyarwanda audio file
Click "Transcribe Audio" to get the text

Option 2: Record Audio

Click on the "Record Audio" tab
Click the microphone button to start recording
Speak in Kinyarwanda
Stop recording and click "Transcribe Recording"

📊 Performance

This model has been optimized for:

Clear speech recognition in various acoustic conditions
Multiple Kinyarwanda dialects and accents
Noise robustness for real-world audio
Fast processing suitable for real-time applications

🤝 About WakandaAI

WakandaAI is dedicated to advancing AI technologies for African languages and communities. This project is part of our mission to make speech recognition accessible in Kinyarwanda.

Built with ❤️ for the Kinyarwanda-speaking community