benchflow-ai/skillsbench/filler-word-processing

[
  {"word": "um", "timestamp": 12.5},
  {"word": "like", "timestamp": 25.3},
  {"word": "you know", "timestamp": 45.8}
]

import json
# Word-specific durations (in seconds)
WORD_DURATIONS = {
    "uh": 0.3,
    "um": 0.4,
    "hum": 0.6,
    "hmm": 0.6,
    "mhm": 0.55,
    "like": 0.3,
    "yeah": 0.35,
    "so": 0.25,
    "well": 0.35,
    "okay": 0.4,
    "basically": 0.55,
    "you know": 0.55,
    "i mean": 0.5,
    "kind of": 0.5,
    "i guess": 0.5,
}
DEFAULT_DURATION = 0.4
def annotations_to_segments(annotations_file, buffer=0.05):
    """
    Convert filler word annotations to (start, end) cut segments.
    Args:
        annotations_file: Path to JSON annotations
        buffer: Small buffer before the word (seconds)
    Returns:
        List of (start, end) tuples representing segments to remove
    """
    with open(annotations_file) as f:
        annotations = json.load(f)
    segments = []
    for ann in annotations:
        word = ann.get('word', '').lower().strip()
        timestamp = ann['timestamp']
        # Use word-specific duration, fall back to default
        word_duration = WORD_DURATIONS.get(word, DEFAULT_DURATION)
        # Cut starts slightly before the word
        start = max(0, timestamp - buffer)
        # Cut ends after word duration
        end = timestamp + word_duration
        segments.append((start, end))
    return segments

def merge_overlapping_segments(segments, min_gap=0.1):
    """
    Merge segments that overlap or are very close together.
    Args:
        segments: List of (start, end) tuples
        min_gap: Minimum gap to keep segments separate
    Returns:
        Merged list of segments
    """
    if not segments:
        return []
    # Sort by start time
    sorted_segs = sorted(segments)
    merged = [sorted_segs[0]]
    for start, end in sorted_segs[1:]:
        prev_start, prev_end = merged[-1]
        # If this segment overlaps or is very close to previous
        if start <= prev_end + min_gap:
            # Extend the previous segment
            merged[-1] = (prev_start, max(prev_end, end))
        else:
            merged.append((start, end))
    return merged

def process_filler_annotations(annotations_file, word_duration=0.4):
    """Full pipeline: load annotations -> create segments -> merge overlaps"""
    # Load and create initial segments
    segments = annotations_to_segments(annotations_file, word_duration)
    # Merge overlapping cuts
    merged = merge_overlapping_segments(segments)
    return merged