Getting Started

Choose a transcription engine: local Whisper, Groq or OpenAI

Transcribe offline for free with local Whisper, or use a Groq or OpenAI key for faster results on long videos.

Captixa turns speech into timed words with one of three engines. You choose in Settings, under Transcription, then Engine.

Local Whisper: free and offline

  1. Open Settings and set the engine to Local Whisper.
  2. Under the offline models, click Download next to a model. Large V3 Turbo (about 1.6 GB) is a good balance; Tiny (about 75 MB) is the quickest to download.
  3. That's it. Nothing leaves your computer, and no key is needed.

Groq: fast, with a free tier

  1. Sign in at console.groq.com.
  2. Open API Keys and choose Create API Key. Copy it straight away; Groq only shows it once.
  3. In Captixa, open Settings, set the engine to Groq, paste the key and save.

OpenAI

Paste an OpenAI API key in Settings and pick OpenAI as the engine. OpenAI bills you directly.

What is sent where

With Groq or OpenAI, only the extracted audio is sent for transcription, never the video. Keys are stored on your computer only, one per provider. Translation and Roman-script conversion use the separate AI model setting (Groq, OpenAI or Gemini).

All tutorials Still stuck? Contact us

Ready to caption your first video?

Every style on the Free plan, 5 videos a day. No account and no card needed.