Speech Accent Classification

# Speech Accent Classification
# Based on "Common Voice: A Massively-Multilingual Speech Corpus" (Ardila et al., LREC 2020)
# Task: Classify speaker accents and assess speech quality from audio recordings

annotation_task_name: "Speech Accent Classification"
task_dir: "."

# Data configuration
data_files:
  - sample-data.json
item_properties:
  id_key: "id"
  text_key: "audio_url"

# Output
output_annotation_dir: "annotation_output/"
output_annotation_format: "json"

# Display layout with audio player and transcription
html_layout: |
  <div class="accent-container" style="font-family: Arial, sans-serif; max-width: 750px; margin: 0 auto;">
    <div class="metadata-bar" style="display: flex; gap: 12px; margin-bottom: 14px; flex-wrap: wrap;">
      <div style="background: #e8eaf6; padding: 7px 14px; border-radius: 8px;">
        <strong>Language:</strong> {{language}}
      </div>
      <div style="background: #e0f2f1; padding: 7px 14px; border-radius: 8px;">
        <strong>Speaker:</strong> {{speaker_id}}
      </div>
      <div style="background: #fff3e0; padding: 7px 14px; border-radius: 8px;">
        <strong>Duration:</strong> {{duration_seconds}}s
      </div>
    </div>
    <div class="audio-section" style="background: #263238; padding: 20px; border-radius: 8px; margin-bottom: 16px; text-align: center;">
      <audio controls style="width: 100%;">
        <source src="{{audio_url}}" type="audio/wav">
        Your browser does not support the audio element.
      </audio>
      <p style="color: #b0bec5; font-size: 12px; margin-top: 10px;">Listen to the full recording before annotating</p>
    </div>
    <div class="transcription-section" style="background: #f5f5f5; padding: 16px; border-radius: 8px; border-left: 5px solid #1976d2;">
      <h4 style="margin-top: 0; color: #1565c0;">Provided Transcription</h4>
      <div style="font-size: 15px; line-height: 1.6; font-style: italic;">{{provided_transcription}}</div>
    </div>
  </div>

# Annotation schemes
annotation_schemes:
  # Accent region classification
  - name: "accent_region"
    description: "What regional accent does the speaker have? Choose the most specific category that applies."
    annotation_type: radio
    labels:
      - name: "north-american"
        tooltip: "US or Canadian English accent (General American, Southern, etc.)"
        key_value: "1"
      - name: "british"
        tooltip: "UK English accent (RP, Cockney, Scottish, Welsh, etc.)"
        key_value: "2"
      - name: "australian"
        tooltip: "Australian or New Zealand English accent"
        key_value: "3"
      - name: "south-asian"
        tooltip: "Indian, Pakistani, Bangladeshi, or Sri Lankan English accent"
        key_value: "4"
      - name: "east-asian"
        tooltip: "Chinese, Japanese, Korean, or Southeast Asian English accent"
        key_value: "5"
      - name: "african"
        tooltip: "English accent from any African country"
        key_value: "6"
      - name: "latin-american"
        tooltip: "English accent influenced by Spanish or Portuguese"
        key_value: "7"
      - name: "middle-eastern"
        tooltip: "English accent influenced by Arabic, Farsi, Turkish, etc."
        key_value: "8"
      - name: "european-non-native"
        tooltip: "English accent influenced by French, German, Russian, etc."
        key_value: "9"
      - name: "other"
        tooltip: "Accent that does not fit the above categories"
        key_value: "0"

  # Accent strength
  - name: "accent_strength"
    description: "How strong is the speaker's accent?"
    annotation_type: radio
    labels:
      - name: "native"
        tooltip: "Speaker sounds like a native English speaker from the identified region"
        key_value: "n"
      - name: "mild-accent"
        tooltip: "Slight non-native features but easily understood"
        key_value: "m"
      - name: "moderate-accent"
        tooltip: "Noticeable accent that does not impede understanding"
        key_value: "d"
      - name: "strong-accent"
        tooltip: "Pronounced accent that occasionally makes comprehension harder"
        key_value: "s"
      - name: "very-strong-accent"
        tooltip: "Very heavy accent that frequently impedes comprehension"
        key_value: "v"

  # Audio quality assessment
  - name: "audio_quality"
    description: "How is the overall audio recording quality?"
    annotation_type: radio
    labels:
      - name: "excellent"
        tooltip: "Clear audio with no background noise or distortion"
      - name: "good"
        tooltip: "Mostly clear with minimal background noise"
      - name: "acceptable"
        tooltip: "Some background noise or minor distortion but speech is audible"
      - name: "poor"
        tooltip: "Significant noise or distortion that makes parts hard to hear"
      - name: "unusable"
        tooltip: "Audio is too noisy, distorted, or quiet to evaluate the speaker"

  # Speech clarity
  - name: "speech_clarity"
    description: "How clearly does the speaker articulate their words?"
    annotation_type: radio
    labels:
      - name: "very-clear"
        tooltip: "Every word is distinctly articulated and easy to understand"
        key_value: "a"
      - name: "clear"
        tooltip: "Most words are clearly spoken with occasional mumbling"
        key_value: "b"
      - name: "somewhat-clear"
        tooltip: "Some words are hard to make out"
        key_value: "c"
      - name: "unclear"
        tooltip: "Many words are mumbled or poorly articulated"
        key_value: "x"
      - name: "unintelligible"
        tooltip: "Cannot understand what the speaker is saying"
        key_value: "z"

  # Transcription correction
  - name: "transcription_correction"
    description: "If the provided transcription contains errors, type the corrected version here. Leave blank if the transcription is correct."
    annotation_type: text

# User configuration
allow_all_users: true

# Task assignment
instances_per_annotator: 50
annotation_per_instance: 2

# Detailed annotation instructions
annotation_instructions: |
  ## Speech Accent Classification

  You are classifying speaker accents from audio recordings and assessing
  speech and audio quality. These recordings come from the Mozilla Common Voice
  corpus of read English speech.

  ### Your Tasks:

  1. **Listen to the full audio clip** at least once before annotating.
  2. **Classify the accent region**: Identify the speaker's regional accent.
  3. **Rate accent strength**: How strong is the accent?
  4. **Assess audio quality**: Rate the recording quality (noise, distortion).
  5. **Rate speech clarity**: How well does the speaker articulate?
  6. **Correct transcription**: Fix any errors in the provided transcription.

  ### Accent Classification Tips:
  - Focus on vowel pronunciation, intonation patterns, and rhythm.
  - "Native" accent strength means the speaker sounds like a native speaker
    from the identified region (not necessarily a native English speaker).
  - If you cannot determine the accent, listen again at a slower speed.
  - For non-native speakers, classify by their L1 influence region.

  ### Audio Quality vs. Speech Clarity:
  - **Audio quality** refers to the recording itself (noise, volume, distortion).
  - **Speech clarity** refers to how the speaker articulates (mumbling, speed).
  - A clear speaker can have a poor recording, and vice versa.

  ### Transcription Correction:
  - Only correct the transcription if you hear clear errors.
  - Leave the correction field blank if the transcription matches the audio.
  - Common errors include misheard words, missing words, or wrong homophones.