Audio Classification and Event Detection with ML Training Course

Foundations of Audio Classification

• Sound event types: environmental, mechanical, and human-generated
• Overview of use cases: surveillance, monitoring, and automation
• Distinguishing between audio classification, detection, and segmentation

Audio Data and Feature Extraction

• Common audio file types and formats
• Considerations for sampling rate, windowing, and frame size
• Extraction of MFCCs, chroma features, and mel-spectrograms

Data Preparation and Annotation

• Utilising UrbanSound8K, ESC-50, and custom datasets
• Labelling sound events and defining temporal boundaries
• Dataset balancing and audio augmentation techniques

Constructing Audio Classification Models

• Application of convolutional neural networks (CNNs) for audio
• Model input structures: raw waveforms versus extracted features
• Loss functions, evaluation metrics, and managing overfitting

Event Detection and Temporal Localisation

• Frame-based and segment-based detection strategies
• Post-processing detections via thresholds and smoothing
• Visualising predictions on audio timelines

Advanced Topics and Real-Time Processing

• Transfer learning for scenarios with limited data
• Model deployment using TensorFlow Lite or ONNX
• Streaming audio processing and latency optimisation

Project Development and Application Scenarios

• Designing a comprehensive pipeline from ingestion to classification
• Developing a proof-of-concept for surveillance, quality control, or monitoring
• Logging, alerting, and integration with dashboards or APIs

Summary and Next Steps