super-transcribe

Route speech-to-text transcription to Parakeet or faster-whisper backends.

Updated Feb 18, 2026

Search Tags:#automation #workflow #translation #subtitles #transcription #speech-to-text #diarization

Ready to run

npx skills add https://github.com/ThePlasmak/super-transcribe --skill super-transcribe

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: super-transcribe
Source: https://github.com/ThePlasmak/super-transcribe/tree/main
Command: npx skills add https://github.com/ThePlasmak/super-transcribe --skill super-transcribe

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires faster-whisper, nemo_toolkit, pyannote.audio, torch, torchaudio, omegaconf, langdetect, yt-dlp, and includes scripts (resource) components.

What problem does it solve?

This Skill provides a single entry point for converting speech to text by automatically selecting the best backend (Parakeet or faster-whisper) and lazily loading dependencies on first use, simplifying setup and improving transcription performance.

Core Features & Use Cases

Unified transcription with automatic routing to the optimal backend for speed and accuracy.
Output formats include text, json, srt, vtt, ass, lrc, ttml, csv, tsv, html; supports diarization and translation.
Use cases include meeting transcripts, podcast subtitling, and multi-language transcription with speaker labeling.

Quick Start

Run a transcription with a single command to generate a transcript using the best available backend.

Frequently Asked Questions about super-transcribe

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I auto-route speech-to-text to the best transcription backend?▼

Auto-routing speech-to-text evaluates available backends like Parakeet and faster-whisper, selecting the optimal one for speed and accuracy. It simplifies setup by lazily loading dependencies on first use and adapting to hardware for efficient transcription.

Can I generate srt and vtt subtitles with speaker diarization?▼

Yes, you can generate srt and vtt subtitles with speaker diarization. The transcription supports multiple output formats including text, json, ass, lrc, ttml, csv, and html, while labeling speakers for meeting transcripts and podcasts.

Does faster-whisper support multi-language translation and transcription?▼

Yes, faster-whisper supports multi-language translation and transcription. The system routes audio to this backend to detect languages and translate speech, adapting to input formats for efficient multi-language processing.

What's the best way to transcribe podcasts without manual dependency setup?▼

The best way to transcribe podcasts without manual setup is using a unified entry point that lazily loads required tools like faster-whisper and nemo_toolkit on first use. This automates environment configuration and routes audio to the best backend.

Do I need a GPU to run pyannote.audio for speaker diarization?▼

No, you do not strictly need a GPU to run pyannote.audio for speaker diarization. The system adapts to available hardware, configuring processing for either GPU or CPU to ensure transcription and diarization still function efficiently.