Too much typing?
Just speak.
打字太累？
用說的才對味。
Hold a hotkey, speak naturally, release. In under 3 seconds AI polishes your voice into clean text — pasted right where your cursor sits. 按住快捷鍵、自然說話、鬆開。3 秒內，AI 把你的語音潤成乾淨文字，直接貼在游標處。
Notch HUD Overlay
Three steps. Speech becomes text. 三個步驟，把你的聲音變成文字。
Not just fast. It's yours. 不只快。還是你的。
Click any feature on the left — we'll show you exactly how it works. 點選左側每個特色 — 我們會實時展示它是怎麼運作的。
Your Keys, Your Data 你的金鑰，你的資料
Plug in your own key from Groq, OpenAI, Gemini, OpenRouter, or NVIDIA NIM — switch any time. Voice data goes straight from your device to the provider you choose, never through our servers. No subscription, no account, no uploads. 自備 Groq、OpenAI、Gemini、OpenRouter 或 NVIDIA NIM 任何一家的 API Key — 隨時切換。你的語音資料直接從電腦送到你選的供應商 — 永遠不經過我們的伺服器。沒有訂閱、沒有帳號、沒有上傳。
100% Open Source · Free Forever 100% 開源 · 永久免費
MIT licensed, community-built. Every core path — speech, AI polish, vocabulary sync — is on GitHub. Read it, modify it, ship it. No paywall, no tracking. MIT 授權、社群共建。所有核心邏輯都在 GitHub 上 — 語音處理、AI 潤稿、詞彙同步。你可以閱讀、改造、重新發佈。沒有付費牆、沒有追蹤。
Smart Vocabulary 智慧詞彙庫
The AI learns from every correction you make. Paste a URL to import domain terms. Synced across mac / Windows / Android through Google Drive — open a new device and it already knows you. 你每一次修正，AI 都在學。貼上一段網址就能匯入專業術語。透過 Google Drive 在 mac / Windows / Android 之間同步 — 打開新裝置就是熟悉的你。
Context-Aware Tone 感知你的情境
TypeLate knows what you're writing in. Formal tone in Mail, casual in iMessage, symbols preserved in VS Code — no settings to flip. TypeLate 知道你在寫什麼。在 Mail 裡用正式語氣，在 iMessage 裡口語化，在 VS Code 裡保留程式符號 — 不用切換設定。
Faster Than Thinking 快如思考
Using Groq's Whisper API — from releasing the hotkey to text landing averages under 3 seconds. Swap to OpenAI Whisper or a local model any time. 使用 Groq 的 Whisper API，從鬆開快捷鍵到文字落地 — 平均低於 3 秒。可隨時切換到 OpenAI Whisper 或本地模型。
Wherever you type,
you can speak. 能打字的地方，
都能用說的。
TypeLate works at the OS level. No extensions, no browser plugins. Anywhere you have a cursor, you can speak. TypeLate 運作在作業系統層級，沒有任何外掛、沒有瀏覽器擴充。有游標的地方，就能說話。
How We Compare 功能比較
The only open-source voice-to-text tool with AI post-processing. 唯一具備 AI 後處理的開源語音轉文字工具。
TypeLate ← That's us
Wispr Flow
Typeless
macOS Dictation
Ready to stop typing? 準備好用說的了嗎？
Coming Soon — vote on GitHub to help us prioritize 即將推出 — 到 GitHub 許願，幫助我們排定優先順序
Built by the community 由社群共同打造
Free and open source. Always. 免費且開源，永遠。
Frequently Asked Questions 常見問題
Is my API key safe? 我的 API Key 安全嗎？
Yes. Your API key is stored locally on your machine and never sent to our servers. Voice data goes directly from your microphone to the AI provider. TypeLate is open source — you can verify this yourself. 是的。你的 API Key 只儲存在你的電腦上，永遠不會傳送到我們的伺服器。語音資料從麥克風直接傳送給 AI 供應商。TypeLate 是開源的 — 你可以自行驗證。
Is my voice data private? 我的語音資料隱私嗎？
Completely. Audio is sent directly to the AI provider you configure (Groq by default). We have zero telemetry servers and collect no data. The audio is never stored — it's processed in real-time and discarded. 完全隱私。音訊直接傳送到你設定的 AI 供應商（預設為 Groq）。我們沒有遙測伺服器，也不收集任何資料。音訊不會被儲存 — 即時處理後立即丟棄。
Which languages are supported? 支援哪些語言？
Whisper supports 50+ languages for transcription. The app interface is available in English, 繁體中文, 简体中文, 日本語, and 한국어. You can speak in any Whisper-supported language and get accurate results. Whisper 支援 50+ 種語言的語音轉錄。App 介面支援英文、繁體中文、简体中文、日本語、한국어。你可以用任何 Whisper 支援的語言說話，都能得到準確結果。
How is this different from macOS Dictation? 這和 macOS 內建聽寫有什麼不同？
TypeLate adds AI post-processing that macOS Dictation doesn't have — it removes filler words, fixes grammar, adjusts tone based on context, and applies your custom prompts. It also supports custom vocabulary and cloud sync. TypeLate 提供 macOS 聽寫沒有的 AI 後處理 — 去除贅詞、修正文法、根據情境調整語氣，還能套用你的自訂 Prompt。同時支援自訂詞彙庫和雲端同步。
Is TypeLate free? TypeLate 免費嗎？
Yes, TypeLate is completely free and open source under the MIT license. You only pay for the AI API usage (Groq offers a generous free tier). No subscriptions, no premium features behind a paywall. 是的，TypeLate 完全免費且以 MIT 授權開源。你只需支付 AI API 的使用費用（Groq 提供慷慨的免費額度）。沒有訂閱制，沒有付費牆。
Does it work offline? 可以離線使用嗎？
Currently, TypeLate requires an internet connection for AI transcription and post-processing. Offline support with local models is on our roadmap. 目前 TypeLate 需要網路連線進行 AI 語音轉錄和後處理。本地模型的離線支援已在我們的開發路線圖中。
Which AI providers are supported? 支援哪些 AI 供應商？
Groq and OpenAI are both supported for Whisper transcription and LLM post-processing. Switch providers anytime in Settings. Support for Azure and local models is planned. Groq 和 OpenAI 皆可用於 Whisper 語音轉錄和 LLM 後處理。可在設定中隨時切換供應商。Azure 和本地模型的支援已在規劃中。