On the recordRead 12 Oct 202622 facts
asub
asub · Italy
On the record
- Batch processing
- It can process supported media files in a folder, loading the Whisper model once for the batch.github.com · read 12 Oct 2026
- Hardware
- It runs on CPU or GPU and supports int8 quantisation for low-memory setups.github.com · read 12 Oct 2026
- Hardware and models
- It supports CPU and GPU processing, offers Whisper model choices from tiny to large-v3, and auto-detects device and compute type when omitted.github.com · read 12 Oct 2026
- Installation status
- The README labels PyPI installation as “once published.”github.com · read 12 Oct 2026
- Integrations
- The project uses faster-whisper, deep-translator with Google Translate, Silero VAD, and optional WhisperX and pyannote diarization models.github.com · read 12 Oct 2026
- Language requirements
- The package metadata requires Python 3.10 or later.github.com · read 12 Oct 2026
- Languages
- It automatically detects the source language or lets users specify it, and translates subtitles into 100+ languages via Google Translate without an API key.github.com · read 12 Oct 2026
- License
- The repository identifies its license as MIT.github.com · read 12 Oct 2026
- Limits
- Folder processing scans only the top level, and translation batches can encounter network or rate-limit issues.github.com · read 12 Oct 2026
- Other operating systems
- The README gives Linux and macOS shell instructions for setting the Hugging Face token, and the package metadata classifies the software as OS independent.github.com · read 12 Oct 2026
- Output formats
- It can write subtitles in SRT and WebVTT formats.github.com · read 12 Oct 2026
- Purpose
- asub generates and translates subtitles from audio or video files using Whisper.github.com · read 12 Oct 2026
- Python API
- The README provides a Python API for loading a model, transcribing media, translating segments, and writing subtitle files.github.com · read 12 Oct 2026
- Speaker identification
- Yesgithub.com · read 22 Sep 2026
- Speaker labels
- Optional WhisperX-powered diarization adds anonymous speaker labels, and the README notes that overlapping speech can be difficult to handle.github.com · read 12 Oct 2026
- Subtitle export
- srt and vttgithub.com · read 22 Sep 2026
- Subtitle formats
- It outputs SRT and WebVTT subtitle files.github.com · read 12 Oct 2026
- Transcription
- It uses faster-whisper and offers automatic language detection or manual source-language selection.github.com · read 12 Oct 2026
