Skip to main content

Project specified Dataset Manager (Audio / Gesture)

Audio Classification

Gesture Recognition

The Local Dataset Manager is the project-specific workspace that feeds data directly into the feature generation and training pipeline.

Transform Data Tab — Audio Conversion

Transform Data tab — Audio Conversion

  1. Add Files or Add Folder to load audio files
  2. Click Analyze to read metadata (format, size, duration)
  3. Click Convert to WAV — originals are never overwritten
  4. Click Export ZIP to download all WAV files

Transform Data Tab — CSV Normalisation (Gesture / Time-Series)

  1. Add CSV files via Add Files or Add Folder
  2. Files load with Pending status
  3. Click Analyze to process the current page (50 files per page); navigate Prev/Next
  4. Click Normalize Data to apply [-1, +1] min-max normalisation
  5. Click Export ZIP (only enabled after all pages are analysed)
IMPORTANT

Export ZIP is only available after every page is analysed. Original source files are never modified.

StatusMeaning
PendingLoaded but not yet analyzed
ReadyValues within [-128, 255]
Out of RangeValues outside [-128, 255] — normalisation recommended
NormalizedSuccessfully normalised to [-1, +1]
No numeric colsNo analysable numeric columns
ErrorFile could not be processed

Transform Data tab — image editing