Music Tagging

Multi-label music tagging following MagnaTagATune dataset format for instrument and genre annotation.

Configuration Fileconfig.yaml

annotation_task_name: "Music Tagging"

port: 8000

# Data configuration
data_files:
  - "data/music_clips.json"

item_properties:
  id_key: id
  text_key: text

# Annotation schemes
annotation_schemes:
  # Instruments
  - annotation_type: multiselect
    name: instruments
    description: "What instruments can you hear? (Select all)"
    labels:
      - Guitar (acoustic)
      - Guitar (electric)
      - Piano/keyboard
      - Drums/percussion
      - Bass
      - Violin/strings
      - Brass (trumpet, etc.)
      - Woodwind (flute, sax, etc.)
      - Synthesizer
      - No instruments

  # Vocals
  - annotation_type: multiselect
    name: vocals
    description: "Describe the vocals (if present)"
    labels:
      - Male vocals
      - Female vocals
      - Choir/group
      - Rapping/spoken word
      - Instrumental only (no vocals)

  # Genre tags
  - annotation_type: multiselect
    name: genre
    description: "What genre(s) does this fit? (Select up to 3)"
    labels:
      - Rock
      - Pop
      - Electronic/dance
      - Hip-hop/rap
      - Jazz
      - Classical
      - Folk/country
      - R&B/soul
      - Metal
      - Blues
      - World/ethnic
      - Ambient

  # Mood/emotion tags
  - annotation_type: multiselect
    name: mood
    description: "What mood/feeling does this evoke? (Select up to 3)"
    labels:
      - Happy/upbeat
      - Sad/melancholic
      - Energetic/exciting
      - Calm/relaxing
      - Aggressive/intense
      - Romantic
      - Dark/mysterious
      - Nostalgic

  # Acoustic features
  - annotation_type: radio
    name: tempo
    description: "How would you describe the tempo?"
    labels:
      - Very slow
      - Slow
      - Medium
      - Fast
      - Very fast

  # Production quality
  - annotation_type: likert
    name: quality
    description: "Rate the production/recording quality"
    size: 5
    min_label: "Low quality"
    max_label: "High quality"

  # Familiarity
  - annotation_type: radio
    name: familiarity
    description: "Do you recognize this song?"
    labels:
      - Yes, I know this song
      - Sounds familiar
      - Never heard it before

# User settings
allow_all_users: true
instances_per_annotator: 200

# Output
output_annotation_dir: "annotation_output/"
output_annotation_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 music-tagging
cd music-tagging
# Copy config.yaml from above
potato start config.yaml

Details

Annotation Types

multiselectlikert

Domain

Audio

Use Cases

music tagginggenre classificationinstrument recognition

Related Designs

Audio Transcription Review

Review and correct automatic speech recognition transcriptions with waveform visualization.

likertmultiselect

Acoustic Scene Classification

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

radiolikert

AnnoMI Counselling Dialogue Annotation

Annotation of motivational interviewing counselling dialogues based on the AnnoMI dataset. Annotators label therapist and client utterances for MI techniques (open questions, reflections, affirmations) and client change talk (sustain talk, change talk), with quality ratings for therapeutic interactions.

radiomultiselect