Mispher

Dictate, rewrite, translate, and run a local agent, all on your Mac

Overview

Mispher is a Mac dictation app that goes further than transcription. Speak, and you can choose what happens next: get the raw text, get it rewritten, get it translated, or hand it to a local agent that goes and does something with it, like filing a note or running a small task through an MCP-connected tool. All the transcription and language models run on-device using Apple Silicon, so nothing gets sent to a server, and there's no account to create.

It's free and open source under the MIT license, with the code on GitHub, which is unusual for a tool this capable. The tradeoff is the requirement: Apple Silicon and macOS 26, so it won't run on Intel Macs or older OS versions. If you're on a supported Mac and want dictation with real editing and light agent power without a subscription or cloud dependency, it's worth installing. If privacy isn't a priority for you, cloud-based dictation tools may offer more polish and support.

Key Features

  • On-Device Transcription: Multiple transcription models including Parakeet, Nemotron, and Qwen3-ASR covering roughly 40 languages.

  • Rewrite And Translate: Select text and rewrite it or translate it on the spot using a local language model.

  • Local Agent Mode: An on-device LFM2.5 agent plans and calls tools against Apple Notes, clipboard, files, and MCP servers.

  • Tool Approval Policies: Every tool and MCP server can be set to auto-approve, ask first, or deny, so the agent never runs something silently.

  • Fully Customizable: Remap hotkeys and the input dial, choose from three HUD styles, and swap models or prompts.

  • No Cloud, No Account: Nothing leaves the device, and there's no sign-up required to use it.

Pricing

Starting price

Free (open source, MIT license)

  • Free And Open Source: $0. MIT license, full feature set, no account or subscription required.

Disclaimer: pricing may change, confirm on Mispher's own site or GitHub page before installing.

Pros

  • Completely Free: Open source under MIT, no subscription, no paywall.

  • Fully On-Device: No cloud calls, no account, strong privacy by default.

  • More Than Dictation: Rewrite, translate, and agentic actions in one tool instead of three separate apps.

  • Granular Tool Control: Per-tool approval policies keep the local agent from acting without your say-so.

Cons

  • Strict Hardware Requirement: Needs Apple Silicon and macOS 26, so Intel Macs and older systems can't run it.

  • No Official Support Channel: As an open-source, community-run project, there's no guaranteed support or roadmap.

  • Mac Only: No Windows or mobile version, so it's not usable outside the Apple desktop ecosystem.

What Makes It Unique

  • Dictate-To-Agent Pipeline: One voice command can go straight from transcription to a local AI agent taking action on your files or notes, with approval gates at every tool call.

Kay Score

7

/ 10

Tool Information

Pricing

Free (open source, MIT license)

Category

Voice & Video

Platform

Web / iOS / Android

Last Updated

Top Alternatives

Synthesia

Studio-quality AI avatar videos in 160+ languages, no camera or studio needed.

Synthesia turns a script into a professional talking-avatar video in minutes. Pick from 240+ AI avatars and 1,000+ voices, and it handles translation into 160+ languages automatically. No filming, no editing software.

Voice & Video

$18/mo

Synthesia

Studio-quality AI avatar videos in 160+ languages, no camera or studio needed.

Synthesia turns a script into a professional talking-avatar video in minutes. Pick from 240+ AI avatars and 1,000+ voices, and it handles translation into 160+ languages automatically. No filming, no editing software.

Voice & Video

$18/mo

Kling AI

Cinematic AI video generation from a single text or image prompt.

Kling AI is a next-generation AI creative studio built on a fully upgraded multimodal architecture, enabling anyone to generate cinematic-quality videos, images, and audio from simple text or image prompts. With its powerful Kling 3.0 model series at its core, it delivers exceptional consistency across complex multi-scene storytelling — making it one of the most advanced AI video generation platforms available today.

Voice & Video

$10/mo

Kling AI

Cinematic AI video generation from a single text or image prompt.

Kling AI is a next-generation AI creative studio built on a fully upgraded multimodal architecture, enabling anyone to generate cinematic-quality videos, images, and audio from simple text or image prompts. With its powerful Kling 3.0 model series at its core, it delivers exceptional consistency across complex multi-scene storytelling — making it one of the most advanced AI video generation platforms available today.

Voice & Video

$10/mo

Heygen

Studio-quality talking-avatar videos without a camera, crew, or editing skills.

HeyGen is an AI-powered video generation platform that lets anyone create professional, studio-quality videos using lifelike AI avatars — no camera, crew, or editing skills required. With 230+ avatars across 140+ languages and features like digital twins, real-time avatar interaction, and AI-powered scripting, HeyGen is redefining how businesses, marketers, and creators produce video content at scale

Voice & Video

$29/mo

Heygen

Studio-quality talking-avatar videos without a camera, crew, or editing skills.

HeyGen is an AI-powered video generation platform that lets anyone create professional, studio-quality videos using lifelike AI avatars — no camera, crew, or editing skills required. With 230+ avatars across 140+ languages and features like digital twins, real-time avatar interaction, and AI-powered scripting, HeyGen is redefining how businesses, marketers, and creators produce video content at scale

Voice & Video

$29/mo

Never miss an AI breakthrough

Join 10,000+ subscribers getting the latest AI tools, news, and tips delivered straight to their inbox every Tuesday.