scene-otsu 0.1.5

SRT subtitle scene splitter using Otsu method with OpenAI embeddings

scene-otsu

A Python library for scene splitting SRT subtitle files using the Otsu method. Uses OpenAI's embedding models to find semantically appropriate scene boundaries.

Features

• SRT Subtitle Parsing: Parse SRT format subtitle files
• Otsu Method Scene Splitting: Recursively apply multi-dimensional Otsu method to detect semantically appropriate scene boundaries
• OpenAI Embeddings: Use OpenAI's embedding models to calculate semantic similarity of text
• Token Limit Support: Split scenes based on specified maximum token count

Installation

pip install scene-otsu

Usage

Basic Example

from scene_otsu import SceneSplitter

# Set OpenAI API key
api_key = "your-openai-api-key"

# Initialize SceneSplitter
splitter = SceneSplitter(api_key=api_key)

# Process SRT string
srt_content = """
1
00:00:00,000 --> 00:00:05,000
Hello, welcome to this video.

2
00:00:05,000 --> 00:00:10,000
Today we will discuss machine learning.

3
00:00:10,000 --> 00:00:15,000
Let's start with the basics.
"""

# Execute scene splitting (max 200 tokens)
result = splitter.process(srt_content, max_tokens=200)
print(result)

Using SubtitleParser

from scene_otsu import SubtitleParser

# Parse SRT string
srt_content = """
1
00:00:00,000 --> 00:00:05,000
First subtitle

2
00:00:05,000 --> 00:00:10,000
Second subtitle
"""

# Basic parsing
subtitles = SubtitleParser.parse_srt_string(srt_content)
# Returns: [("00:00:00,000", "00:00:05,000", "First subtitle"), ...]

# Parse as scene information
scenes = SubtitleParser.parse_srt_scenes(srt_content)
# Returns: [{"index": 1, "start_time": "00:00:00,000", "end_time": "00:00:05,000", ...}, ...]

API Reference

SceneSplitter

__init__(api_key: str, model: str = "text-embedding-3-small", batch_size: int = 16)

Initialize SceneSplitter.

Parameters:

• api_key: OpenAI API key
• model: OpenAI embedding model to use (default: "text-embedding-3-small")
• batch_size: Batch size for embedding generation (default: 16)

process(srt_string: str, max_tokens: int = 200) -> str

Process SRT string and return scene-split SRT string.

Parameters:

• srt_string: Input SRT string
• max_tokens: Maximum tokens per chunk (default: 200)

Returns:

• Scene-split SRT string

SubtitleParser

parse_srt_string(srt_string: str) -> List[Tuple[str, str, str]]

Parse SRT string.

Returns:

• List in the format [(start_timestamp, end_timestamp, text), ...]

parse_srt_scenes(srt_string: str) -> List[Dict[str, Any]]

Convert SRT to scene-based dictionary.

Returns:

• List in the format [{index, start_time, end_time, start_sec, end_sec, text}]

parse_timestamp(timestamp: str) -> float

Convert timestamp string to seconds.

Parameters:

• timestamp: Timestamp string (e.g., "00:00:05,000")

Returns:

• Seconds (float, including milliseconds)

Requirements

• Python 3.11 or higher
• OpenAI API key

Dependencies

• numpy >= 2.3.5
• openai >= 2.8.1
• scikit-learn >= 1.7.2
• tiktoken >= 0.12.0
• tqdm >= 4.67.1

License

MIT License

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Author

Yuki Harada