/ASR_PROJECTS

A portfolio of all my automatic speech recognition and audio classification projects

Primary LanguageJupyter NotebookApache License 2.0Apache-2.0

AUTOMATIC SPEECH RECOGNITION (ASR) PROJECTS

Overview

Welcome to my Automatic Speech Recognition (ASR) Portfolio, where I showcase a series of projects focused on advancing voice technology, particularly in African languages. As an AI engineer with a profound passion for machine learning and deep learning, I have immersed myself in mastering the intricacies of ASR frameworks, ensuring each project reflects a dedication to efficiency, fundamental principles, and cutting-edge solutions.

Explore the diverse projects presented here, each designed to demonstrate the application of ASR technology in solving unique challenges related to voice recognition and classification. Whether it's building models for Swahili audio classification or training ASR models for Kiswahili, these projects underscore my commitment to pushing the boundaries of voice technology.

Projects

  1. Swahili Audio Classification Challenge

    • Overview: Developing an automatic speech recognition solution to classify simple Swahili audio into text.
    • Aim: Providing innovative solutions for telecommunication companies, translation services, public health, and emergency services.
    • Technologies: Python, Fastai, PyTorch.
  2. Mozilla Common Voice Hackathon I Nairobi - Kiswahili ASR Model Training

    • Overview: Training an ASR model for Kiswahili using the Mozilla Common Voice (MCV) dataset.
    • Aim: Fostering a developer community equipped with skills to leverage voice technology and solve real-world problems.
    • Importance: Contributing to inclusive voice technology and problem-solving in East Africa.
    • Technologies: ASR frameworks (e.g., PyTorch, Wav2vec, conformer-ctc, whisper), open-source tools.
  3. Noise Data Classification

    • Overview: Training a noise classification model to categorize different types of noise.
    • Aim: Generating data-driven insights to address and mitigate the impact of noise pollution.
    • Importance: Improving urban well-being, empowering citizens, and developing data-driven solutions.
    • Technologies: Python, Fastai, PyTorch.

Future Developments

Continuing my exploration of innovative solutions in the realm of ASR, I am committed to advancing voice technology for impactful applications. Stay tuned for updates and new projects that push the boundaries of what's possible in the field of automatic speech recognition.

Thank you for visiting my ASR Portfolio!