Skip to content

Dictionary

The dictionary helps Voquill handle domain-specific vocabulary, names, and acronyms that speech recognition might otherwise get wrong. It has two features: glossary terms and replacement rules.

Glossary terms are fed directly to the transcription model as hints. When you add a glossary term, the speech recognition model uses it as a reference during transcription itself, making it more likely to produce the correct word.

This is especially useful for:

  • Product names (e.g., “Voquill”, “PostgreSQL”)
  • People’s names (e.g., “Josiah”, “Satya Nadella”)
  • Technical jargon (e.g., “Kubernetes”, “WebSocket”)
  • Anything the transcription model consistently mishears

Replacement rules find and replace text in the raw transcript, right before post-processing runs. If Voquill keeps getting a word wrong, you can add a replacement rule to fix it automatically.

For example:

InputOutput
GPTChatGPT
V quillVoquill

You can also use replacement rules as snippets. Define a short trigger phrase and have it expand into something longer — useful for boilerplate text, signatures, or common phrases you dictate frequently.

InputOutput
sig blockBest regards, Josiah — Voquill Team
addr123 Main Street, Suite 400, New York, NY

Since replacements happen on the raw transcript before the AI post-processing step, the language model sees the corrected text and works from there.

Open the dictionary from settings. From there you can add, edit, and delete both glossary terms and replacement rules. Changes take effect immediately on your next transcription.