Episode

Ep 55 | Audio analytics with Azure Automated ML

with Seth Juarez, Serge Retkowsky

AI Show

Serge Retkowsky is back to demonstrate how audio processing can consist of extracting audio signal information into spectrograms (time vs frequency vs Db) images that we can use to build a custom vision model with Azure. We'll also extract some audio components and use a generic classification model with Azure ML and its AutoML features.

Chapters

  • 00:00 - Welcome to the AI Show
  • 01:27 - Intro to Audio analytics with Azure Automated ML
  • 03:02 - Fast Fourier Transform (FFT) - Time Domain Signal to Frequency Domain Signal
  • 05:10 - Spectogram of how the spectrum of of frequencies vary over time
  • 07:50 - Demo - Music Genre Classification
  • 13:15 - Creating a Chromagram
  • 14:05 - Spectograms generation
  • 15:04 - Music Genre Classification with Automated ML for Images
  • 16:34 - Calling AutoML CV Model
  • 22:42 - Demo 2
  • 25:29 - Learn more