Amical

Amical is an open-source AI dictation app for Mac, Windows, iOS and Android that turns speech into text formatted for the app in use. It supports local or cloud models, custom vocabulary, shortcuts and 100+ languages.

Amical is a voice dictation app that converts speech into formatted text and types it into whichever application has focus. It runs on Mac, Windows, iOS and Android, and the core desktop app is open source, with the code published on GitHub. The homepage positions it as a faster way to write emails, chat messages, code prompts and notes, claiming roughly four times the speed of typing. The main differentiator is context awareness. Rather than dumping raw transcription, Amical looks at the app being used and adjusts tone and structure: a professional register for Gmail, a looser one for Instagram or Discord. It also corrects grammar, fixes pronouns, can add contextual emojis, and tries to match the user's own vocabulary and writing style. Dedicated pages show the behavior in Gmail, Slack, Cursor, Notion, Jira, ChatGPT, Discord and WhatsApp. In practice, the user installs the app, sets a keyboard shortcut, holds or toggles it while speaking, and the cleaned-up text appears at the cursor. A custom vocabulary list covers project names, jargon and proper nouns that general speech models tend to mangle. Custom shortcuts support hands-free voice commands, and dictation works in more than 100 languages, including mixed-language speech. Model choice is another notable point. Users can pick cloud-based models or run everything locally for privacy, switching between providers such as Ollama, OpenAI and OpenRouter, or using Amical's own option. Reviews quoted on the site highlight that the local mode works offline and costs nothing. Amical fits between the basic built-in dictation of macOS and Windows and paid commercial dictation tools such as Aqua Voice or Typeless, which users on the site mention as alternatives. It suits developers, knowledge workers, multilingual writers and privacy-conscious users who want AI-polished dictation without sending audio to a third party by default.