speakmore

SpeakMore:口述成文,划词即译。开源、免费的语音输入与翻译工具,支持 Mac 与 Windows:语音输入、语音翻译、划词翻译。兼容 Gemini、OpenAI、通义千问、DeepSeek、Kimi

What is speakmore?

SpeakMore:口述成文,划词即译。开源、免费的语音输入与翻译工具,支持 Mac 与 Windows:语音输入、语音翻译、划词翻译。兼容 Gemini、OpenAI、通义千问、DeepSeek、Kimi、智谱 GLM、MiniMax 及任意 OpenAI 兼容服务,语音输入亦可完全在本机运行。附 Typeless 逆向分析。

Overview of speakmore

What it is

SpeakMore is an open-source, free speech input and translation tool for macOS and Windows. It supports speech input, speech translation, and selection translation, each triggered by a hotkey. It is compatible with major large language models (Gemini, OpenAI, Tongyi Qianwen, DeepSeek, Kimi, Zhipu GLM, MiniMax, and any OpenAI-compatible service). Speech input can also run entirely locally.

Key features

  • Three functions: speech input (right ⌥/right Alt), speech translation (right ⌥ + Space), selection translation (right ⌥ + Enter).
  • Speech input: automatic removal of filler words, repetition, and self-corrections; auto punctuation; longer content split into paragraphs/items; dictionary support for proper nouns and terms.
  • Speech translation: speak in one language, output in another; supports 16 target languages; median model time 2.6s (direct) / 1.3s (mixed).
  • Selection translation: select text in any app, press hotkey, translation appears in a card below the selection; no plugin needed; supports up to 4000 characters.
  • Three recognition modes: direct (audio to text via audio model), mixed (speech-to-text then LLM), offline (local model, no data upload).
  • Model freedom: 29 speech models, separate text model for selection translation; supports OpenAI-compatible services.
  • Hardware freedom: works on Mac and Windows; local models run via Metal, Vulkan, or CPU.
  • Privacy: speech input can be fully offline, no key or account needed.
  • Cost: software is free; estimated monthly cost for recommended speech input combination ¥0–35 (100 uses/day), selection translation ~¥2.4.

Use cases

  • Dictation: speak to write in any application, with automatic cleanup.
  • Real-time translation in chat: speak Chinese, output English directly into the input box for communication with foreign colleagues or clients.
  • Writing emails and commit messages in another language.
  • Translating selected text in PDFs, terminals, web pages, documents, and chat apps without installing plugins.

Getting started

macOS: macOS 12+. Open the dmg, drag SpeakMore to Applications. The package is Developer ID signed and notarized by Apple. Windows: Windows 10 (1809+) or Windows 11, 64-bit. Run the installer; if Windows protected your PC, click 'More info' and 'Run anyway'. Installs for current user only, no admin rights needed. Terminal installation (includes speakmore CLI): bash # macOS curl -fsSL https://dl.zephyrxiang.com/speakmore/install.sh | bash powershell # Windows (PowerShell) irm https://dl.zephyrxiang.com/speakmore/install.ps1 | iex

License and pricing

License: MIT. The software is free and open-source. API usage costs may apply depending on the chosen model and service.

Based on the project's public README and repository metadata.

Add this badge to your README

Featured on YunoDB Badge styles & help

Maintain this project? Show that it is featured on YunoDB.

Markdown[![Featured on YunoDB](https://yunodb.com/badge/zephyr4123speakmore.svg)](https://yunodb.com/tools/zephyr4123speakmore?utm_source=github&utm_medium=badge)
HTML<a href="https://yunodb.com/tools/zephyr4123speakmore?utm_source=github&utm_medium=badge"><img src="https://yunodb.com/badge/zephyr4123speakmore.svg" alt="Featured on YunoDB" height="20"></a>