Launching today
Amical
Dictate hands-free, transcribe meetings, and capture notes.
1 follower
Dictate hands-free, transcribe meetings, and capture notes.
1 follower
Open source AI dictation that detects your active app, formats speech to match, runs locally on Whisper, supports 100+ languages, custom vocab, and works on Mac and Windows.
Amical is an open source AI dictation app that runs locally and adapts its output to whatever application you're typing in.
I'm hunting it because most people don't realise how broken native OS dictation still is in 2025, and this is the most technically honest fix I've seen.
The problem:
Mac and Windows dictation treats every text field the same.
It doesn't know if you're drafting a Jira ticket, a WhatsApp message, or a Slack reply.
The output is raw, context-free, and often wrong on jargon.
The solution:
Amical detects your active application and formats speech accordingly.
Professional tone for Gmail, casual for messaging apps, code-aware for your IDE.
Powered by Whisper for transcription and local LLMs for intelligent processing, no audio leaves your machine by default.
Features worth knowing:
🎙️ Context-aware formatting means you stop manually fixing what native dictation gets wrong
🔐 Local-first means your voice data doesn't touch a third-party server
🌍 100+ language support with custom vocabulary for jargon, names, and technical terms
⌨️ Custom hotkeys and a floating widget mean it stays out of your way until you need it
☁️ Cloud model option (OpenAI, OpenRouter) for when speed beats privacy
Who it's for:
Developers, writers, and knowledge workers who dictate frequently and are frustrated by context-blind transcription. Especially strong for anyone already in the OSS/local-AI camp.
I'm curious: do you currently use any dictation tool in your daily workflow, and what's the one thing that made you stop trusting it?