auto-subs 0.6.2

A powerful, local-first library and CLI for video transcription and subtitle generation using Whisper.

Auto-Subs

Effortless Subtitle Generation from Whisper Transcriptions.

A powerful, local-first library and CLI for generating and editing subtitles with precise, word-level accuracy.

---

Auto-Subs bridges the gap between raw transcription data and perfectly formatted subtitles. Whether you're a developer integrating transcription into your application or a content creator needing quick subtitles, auto-subs provides a robust, simple, and reliable solution.

Key Features

• 🚀 End-to-End Transcription: Go from an audio or video file directly to perfectly timed subtitles in one command.
• 🔥 Hardsubbing (Video Burning): Burn generated subtitles directly into a video file with a simple --burn flag.
• 📝 Rich Programmatic Editing: A powerful, in-memory API to programmatically edit subtitles—shift timing, adjust duration, merge/split segments, and more.
• 🔄 Versatile Format Conversion: Easily convert existing subtitle files between supported formats.
• 🧠 Intelligent Word Segmentation: Automatically splits word-level transcriptions into perfectly timed subtitle lines based on character limits and natural punctuation breaks.
• 📄 Multiple Formats: Generate and convert subtitles in the most popular formats: SRT, VTT, and ASS.
• 🎤 Karaoke-Style Highlighting: Generate word-by-word highlighting ({\k...}) for .ass files, perfect for music videos or language learning.
• 🛡️ Robust Validation: Automatically handles common data issues, like inverted timestamps (start > end), ensuring your process never breaks on imperfect data.
• ⚙️ Simple & Powerful API: Use it as a library with a clean, dictionary-based input that requires no complex objects, or as a feature-rich command-line tool.

Installation

# For subtitle generation and conversion
pip install auto-subs

# To include direct transcription and burning capabilities
pip install auto-subs[transcribe]

Hardsubbing requires FFmpeg to be installed and available in your system's PATH.

Quickstart

As a Command-Line Tool (CLI)

auto-subs provides four powerful commands: transcribe, generate, convert, and burn.

Note: Global options like -q (quiet) or -v (verbose) must be placed before the command name (e.g., auto-subs -q transcribe ...).

# 1. Transcribe a video and burn the subtitles directly into a new file
auto-subs transcribe video.mp4 --model small --burn

# 2. Generate a styled ASS file from an existing transcription JSON
auto-subs generate input.json -f ass -o styled.ass --max-chars 42 --karaoke

# 3. Convert an existing SRT file to ASS format
auto-subs convert subtitles.srt -f ass

# 4. Burn an existing subtitle file into a video
auto-subs burn video.mp4 styled.ass -o final_video.mp4

As a Python Library

Integrate auto-subs directly into your application for full control.

import json
from autosubs import generate, transcribe, load

# --- Generate from existing JSON ---
with open("path/to/transcription.json", "r", encoding="utf-8") as f:
    transcription_data = json.load(f)

srt_content = generate(transcription_data, "srt", max_chars=40)
with open("output.srt", "w", encoding="utf-8") as f:
    f.write(srt_content)

# --- Transcribe directly from a media file ---
try:
    vtt_content = transcribe("path/to/video.mp4", "vtt", model_name="base")
    with open("output.vtt", "w", encoding="utf-8") as f:
        f.write(vtt_content)
except ImportError:
    print("Transcription requires 'auto-subs[transcribe]' to be installed.")

# --- Load and inspect an existing subtitle file ---
subtitles = load("path/to/existing.srt")
print(f"Loaded {len(subtitles.segments)} subtitle segments.")

Powerful Programmatic Editing

auto-subs provides a rich, object-oriented API for advanced, in-memory subtitle manipulation. Once you load or create subtitles, you can edit them and then generate the final output.

from autosubs import load, to_ass
from autosubs.models import AssSettings, AssStyleSettings

# Load an SRT file and automatically generate word-level timings.
# This "upgrades" a standard SRT to a rich, editable format with precise
# word timestamps, enabling fine-grained edits or karaoke generation.
subs = load("input.srt", generate_word_timings=True)

# Get the first subtitle segment
first_segment = subs.segments[0]

# Perform edits using a fluent, chainable API
first_segment.shift_by(-0.25).set_duration(3.5, anchor="start") # Shift 250ms earlier and set duration to 3.5s

# Merge the second and third segments into one
if len(subs.segments) >= 3:
    subs.merge_segments(1, 2)

# Generate a karaoke-style ASS file from the edited subtitles
ass_settings = AssSettings(highlight_style=AssStyleSettings())
karaoke_ass = to_ass(subs, ass_settings)

with open("output.ass", "w") as f:
    f.write(karaoke_ass)

API Design: Simplicity First

The public API of auto-subs is designed to be as simple as possible. Functions like auto_subs.generate() accept a standard Python dictionary (dict).

This approach was chosen intentionally to:

• Reduce Friction: You can directly use the JSON output from Whisper after loading it, without needing to instantiate our internal Pydantic models.
• Decouple Your Code: Your project doesn't need to depend on our internal data structures, making your code more resilient to future updates.

While the input is simple, auto-subs performs robust internal validation, giving you the best of both worlds: a simple API and the safety of strong data validation.

Contributing

Contributions are welcome! If you find a bug or have a feature request, please open an issue. If you'd like to contribute code, please open a pull request.

License

This project is licensed under the MIT License. See the LICENSE file for details.