The Kikuyu ASR Project is an initiative to develop an Automatic Speech Recognition (ASR) system for the Kikuyu language, a low-resource language spoken in Kenya. This ASR system aims to accurately transcribe spoken Kikuyu into text, making it accessible, searchable, and useful for various applications.
- Data Collection: Gather a diverse and representative dataset of spoken Kikuyu language samples.
- Data Preprocessing: Clean, preprocess, and augment the collected data to create a suitable training dataset.
- Feature Extraction: Extract relevant audio features for effective representation of speech signals.
- Model Selection: Experiment with different ML models to identify the best architecture for Kikuyu ASR.
- Training and Optimization: Train and optimize the selected model(s) for accurate transcription.
- Evaluation: Assess the ASR system's performance using relevant metrics.
- Deployment: Develop a user-friendly interface for accessibility.
- Continuous Improvement: Implement feedback mechanisms for ongoing improvement.
- Python 3