joewongjc/type4me

MacOS/iOS 语音输入法，实时识别、大模型文本优化、全本地存储

Type4Me – macOS Voice Input & AI‑Powered Text Assistant

What it is

  • An open‑source macOS application (macOS 14+ / Sonoma) that lets you dictate text directly into any cursor location. It combines speech‑to‑text with optional AI processing such as polishing, translation, answering questions, or controlling the Mac.

How it works

  • You press a global hot‑key, speak, and the spoken words appear as you talk. The raw transcription can then be passed through a large‑language model (LLM) of your choice (OpenAI, Claude, Gemini, Ollama, etc.) or a locally‑run model for further processing.
  • The app supports two deployment modes:
    1. Cloud mode – uses an external speech‑recognition service (e.g., Volcano Engine / 豆包, Soniox, OpenAI) and sends the text to a remote LLM.
    2. Local mode – runs the SenseVoice + Qwen3‑ASR offline recognizer on Apple‑silicon machines (≈8 GB RAM) and still can call a local LLM via Ollama.
  • All configuration (API keys, chosen recognizer, chosen LLM, prompts, hot‑keys) lives on your machine, is backed up daily, and can be exported.

Key features

Feature What it does
Multiple modes Quick (raw transcription only), Smart (auto‑clean‑up & context‑aware phrasing), Translate (to 18 languages), Ask (voice‑driven Q&A on selected text), Mac Control (e.g., “open Safari”, “set volume to 30”), Polishing, Prompt‑generation, To‑Do creation.
Customizable prompts Each mode runs a user‑editable prompt template; placeholders {text}, {selected}, {clipboard} let you stitch together arbitrary workflows.
Hot‑key driven Bind any mode to a global shortcut; choose “hold‑to‑talk” or “toggle‑on/off”.
Vocabulary / Skill system Install the optional type4me‑vocab‑skill to add custom terms and automatic replacements, improving domain‑specific accuracy.
History & usage dashboard Stores raw and processed results, searchable, exportable as CSV; shows token usage, estimated cost, and latency per model.
Menu‑bar control center Quick view of recording status, switch recognizer, LLM, language, etc.
URL scheme type4me://… commands let you trigger actions from Stream Deck, Shortcuts, Raycast, Alfred, or scripts without stealing focus.
Local backup Automatic 7‑day rotation of settings, vocab, mode definitions, and history.
Open‑source & MIT‑licensed Full source available; builds are Apple‑notarized.

Getting started

  1. Download the appropriate DMG (cloud‑recommended or Apple‑silicon local). Install and grant microphone + accessibility permissions.
  2. Add your API keys for a speech‑recognition provider and an LLM (or install Ollama for a local model).
  3. Choose a mode or create your own by editing the prompt template in the settings.
  4. Press the configured hot‑key, speak, and watch the text appear. Use the menu‑bar or URL scheme to invoke extra actions.

Extensibility

  • Plug‑in vocab skill – a separate repo (type4me‑vocab‑skill) that lets the app learn custom terms via voice commands.
  • Model agnostic – any service exposing an OpenAI‑compatible API can be swapped in, including Chinese providers (豆包, DeepSeek, Kimi, etc.) or local Ollama instances.
  • Automation – URL scheme commands enable integration with existing macOS automation tools.

Why it exists The author found existing voice input tools either expensive, closed‑source, or inflexible (no export, no custom prompts, high latency). Type4Me aims to be free, transparent, and adaptable while still offering AI‑enhanced text handling.

Contributing Issues and pull requests are welcomed. The codebase was largely generated with Claude Code, and contributors are encouraged to follow the CONTRIBUTING.md guide.

License MIT License.

All information above is taken directly from the repository’s README.

Related

  • Project
    TypeWhisper/typewhisper-mac

    TypeWhisper is a macOS app that provides on‑device or cloud speech‑to‑text, global hot‑key dictation, and customizable AI workflows (translation, rewriting, extraction, etc.). It supports many local and cloud ASR engines, a plugin SDK, a local REST API, CLI tools, iCloud sync, and premium meeting‑automation features—all wrapped in a polished UI with visual shortcut editing and widget support.

  • Project
    theJayTea/WritingTools

    Writing Tools is a free, open‑source desktop app for Windows, Linux and macOS that lets you invoke an LLM anywhere on your computer with a hot‑key. It can proofread, rewrite, change tone, translate, summarise web pages or YouTube transcripts, and even chat with the model. You can use cloud APIs (Gemini, OpenAI, Anthropic, etc.) or run local models via Ollama, llama.cpp or Apple‑silicon MLX. The app is privacy‑focused (no telemetry, local storage of API keys) and licensed under GPL‑v3.

  • Project
    Beingpax/VoiceInk

    VoiceInk is a native macOS application that provides private, high-accuracy, offline voice-to-text transcription using local AI models.

  • Project
    kdsz001/typefree

    An AI-powered voice input tool for macOS that converts spoken language into polished, formatted text and provides a floating AI assistant for quick queries.

  • Project
    soniqo/speech-swift

    Speech Swift is an open‑source Swift package that provides on‑device speech AI for Apple Silicon. It bundles many modern ASR, TTS, translation, and voice‑agent models (e.g., Qwen3‑ASR, Whisper‑ASR, Omnilingual ASR, Qwen3‑TTS, CosyVoice, PersonaPlex) built on MLX and Core ML, so you can run inference locally with no cloud calls. Each model is a separate Swift‑Package‑Manager product, making it easy to add only what you need. The repo includes quick‑start code, SwiftUI UI helpers, Homebrew install, and extensive documentation.