transcribe-anything

What problem does it solve?

Manually transcribing audio and video content is slow, expensive, and error-prone, especially for long recordings, files with poor audio quality, or content that requires speaker identification. This skill automates the full transcription workflow from raw media to polished, formatted text.

Core Features & Use Cases

• Multi-Backend Support: Choose from 9+ local and cloud ASR engines (Whisper, WhisperX, Groq, Deepgram, etc.) to balance speed, accuracy, cost, and diarization needs.
• Long File Optimization: Automatically preprocesses audio, skips silence, and chunks files to prevent hallucination cascades and meet API size limits for recordings up to 8+ hours.
• Flexible Output: Generates timestamped markdown transcripts, subtitles (SRT/VTT), JSON with word-level timestamps, or plain text for podcasts, meetings, videos, and interviews.

Quick Start

Use the transcribe-anything skill to generate a formatted markdown transcript with speaker labels for the attached 1-hour team meeting recording.