faster-whisper

Overview

Local speech-to-text using faster-whisper, a CTranslate2 reimplementation of OpenAI's Whisper, for 4-6x faster transcription with identical accuracy. With GPU acceleration, expect ~20x realtime transcription.

Key Features

• Transcribe audio/video files
• Generate subtitles (SRT, VTT, ASS, LRC, TTML)
• Identify speakers (diarization labels)
• Transcribe from URLs (YouTube links and direct audio URLs)
• Batch process files (glob patterns, directories, skip-existing support)

How It Works

This skill uses faster-whisper to transcribe audio files locally, without relying on API costs or online services. It supports 99+ languages, auto-detection, and multilingual transcription.

Use Cases

• Transcribe meetings, interviews, podcasts, lectures, and YouTube videos
• Generate subtitles for broadcast-standard formats
• Identify speakers in audio files
• Transcribe podcast feeds and YouTube links
• Batch process files with ETA shown automatically
• Transcribe audio with specific terms or jargon-heavy content
• Preprocess noisy audio before transcription
• Stream output and clip time ranges
• Search the transcript and detect chapters
• Export speaker audio and spreadsheet output