Skip to content
Voquill Docs

Post-Processing

After your speech is transcribed, Voquill runs a post-processing step that uses a large language model to clean up the raw text.

How It Works

Section titled “How It Works”
  1. You record audio and it gets transcribed to raw text (via local Whisper or the Groq API).
  2. The raw transcript is combined with your active style prompt and any dictionary glossary terms.
  3. This combined prompt is sent to a language model that rewrites the text according to your style.
  4. The cleaned-up result is what you see (and what gets pasted if auto-paste is enabled).

Both the raw transcript and the post-processed version are saved, so you can always see the original.

Post-Processing Modes

Section titled “Post-Processing Modes”

Post-processing follows the same mode as your transcription setting:

  • Local — Not currently supported for post-processing. Raw transcripts are returned as-is.
  • API — Uses your Groq API key to call a language model directly.
  • Cloud — Routes through Voquill’s cloud service.

Disabling Post-Processing

Section titled “Disabling Post-Processing”

If you prefer raw transcriptions without any AI cleanup, you can select the Light style, which makes minimal changes, or disable post-processing entirely in settings.