Estimate None timestamp for Whisper output

estimate_none.py

def estimate_none_timestamps(timestamp_list):
    """
    Estimates missing timestamps in a list of timestamp segments based on the character length of segment times.

    Parameters:
    timestamp_list (list): A list of timestamp segments with text.

    Returns:
    list: The list with estimated missing timestamps.
    """
    total_duration = 0
    total_characters = 0

    for segment in timestamp_list:
        start, end = segment["timestamp"]
        if start is not None and end is not None:
            duration = end - start
            characters = len(segment["text"])
            total_duration += duration
            total_characters += characters

    if total_characters > 0:
        avg_duration_per_char = total_duration / total_characters
    else:
        avg_duration_per_char = 0.1  # Default duration per character (assumed)

    for i, segment in enumerate(timestamp_list):
        start, end = segment["timestamp"]
        characters = len(segment["text"])
        estimated_duration = characters * avg_duration_per_char

        if start is None:
            start = timestamp_list[i - 1]["timestamp"][1] if i > 0 and timestamp_list[i - 1]["timestamp"][1] is not None else 0
            segment["timestamp"] = (start, start + estimated_duration)
        if end is None:
            segment["timestamp"] = (start, start + estimated_duration)
    return timestamp_list