Acoustic Scene Classification

Classify audio recordings by acoustic environment following the TUT/DCASE dataset format.

🎧

audio annotation

Configuration Fileconfig.yaml

task_name: "Acoustic Scene Classification"

# Server configuration
server:
  port: 8000

# Audio settings
audio:
  enabled: true
  display: waveform
  waveform_color: "#8B5CF6"
  progress_color: "#A78BFA"
  speed_control: true

# Data configuration
data_files:
  - path: data/scenes.json
    audio_field: audio_file

# Annotation schemes
annotation_schemes:
  # Primary scene category
  - annotation_type: radio
    name: scene_category
    description: "Select the acoustic scene/environment"
    labels:
      - Airport
      - Bus
      - Metro station
      - Metro (inside train)
      - Park
      - Public square
      - Shopping mall
      - Street (pedestrian)
      - Street (traffic)
      - Tram
    keyboard_shortcuts:
      "Airport": "1"
      "Bus": "2"
      "Metro station": "3"
      "Metro (inside train)": "4"
      "Park": "5"
      "Public square": "6"
      "Shopping mall": "7"
      "Street (pedestrian)": "8"
      "Street (traffic)": "9"
      "Tram": "0"

  # Indoor/outdoor
  - annotation_type: radio
    name: environment_type
    description: "Is this primarily indoor or outdoor?"
    labels:
      - Indoor
      - Outdoor
      - Mixed/transitional
      - Cannot determine

  # Scene clarity
  - annotation_type: radio
    name: scene_clarity
    description: "How clearly identifiable is the scene?"
    labels:
      - Very clear (unmistakable)
      - Clear (confident identification)
      - Ambiguous (could be multiple scenes)
      - Unclear (cannot identify)

  # Secondary sounds
  - annotation_type: multiselect
    name: prominent_sounds
    description: "What sounds are most prominent? (Select up to 3)"
    labels:
      - Human speech/chatter
      - Vehicle noise
      - Footsteps
      - Announcements/PA system
      - Nature sounds (birds, wind)
      - Music
      - Machinery/equipment
      - Silence/quiet

  # Confidence rating
  - annotation_type: likert
    name: confidence
    description: "Confidence in your scene classification"
    size: 5
    min_label: "Low"
    max_label: "High"

# User settings
allow_all_users: true
instances_per_annotator: 200

# Output
output:
  path: annotations/
  format: json

Get This Design

This design is available in our showcase. Copy the configuration below to get started.

Quick start:

# Create your project folder
mkdir acoustic-scene-classification
cd acoustic-scene-classification
# Copy config.yaml from above
potato start config.yaml

Details

Annotation Types

radiolikert

Domain

Audio

Use Cases

scene classificationacoustic environmentcontext awareness

Related Designs

Audio-Visual Sentiment Analysis

Rate sentiment in speech segments following CMU-MOSI and CMU-MOSEI multimodal annotation protocols.

likertradio

Speech Emotion Recognition

Classify emotional content in speech following IEMOCAP and CREMA-D annotation schemes.