Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Input0 — A macOS voice input tool: hold a hotkey to record, release to transcribe locally via STT, refine with LLM, and auto-paste into the active text field.
| Date | Stars |
|---|---|
| 2026-07-24 | 289 |
| 2026-07-25 | 289 |
| 2026-07-28 | 289 |
| 2026-07-30 | 289 |
| 2026-07-31 | 290 |
| 2026-08-06 | 290 |
Today
— stars today
This week
+1 stars this week
This month
— stars this month
Momentum
1.0
growth rate 0.35%/day
<div align="right"> <strong>English</strong> | <a href="README.zh-CN.md">简体中文</a> </div> <div align="center"> <img src="landing-page/logo.png" alt="Input 0 Logo" width="120" /> <h1>Input 0</h1> </div> > 🏗️ This project's AI agent harness is built with [Harness Engineering](https://github.com/10xChengTu/harness-engineering). A macOS voice input tool — hold a hotkey to record, release to get polished text auto-pasted into any input field. Local AI transcription → LLM text optimization → auto-paste. Private, fast, effortless.  ## Features - **Press & Speak** — Hold `Option+Space` (customizable) to record, release to transcribe + optimize + paste. No window switching needed. - **Privacy-First Local STT** — Six AI engines (Whisper, SenseVoice, Paraformer, Moonshine, FireRedASR, Zipformer CTC) run entirely on your Mac via Metal GPU. Audio never leaves your device. - **AI-Powered Polish** — LLM auto-corrects grammar, removes filler words, and structures your text. Built-in technical term correction (e.g. phonetic Chinese → "React"), with custom vocabulary support. - **Auto-Paste Anywhere** — Optimized text is automatically pasted into your active input field — Slack, WeChat, VS Code, browsers, any app. - **99+ Languages** — 6 engines, 12 models, covering 99+ languages. The system recommends the best model based on your language. - **On-Demand Models** — Lightweight app, download only the STT models you need. One-click switch, progress display, smart recommendations. - **ESC to Cancel** — Cancel at any stage (recording, transcribing, optimizing) by pressing ESC. - **History** — Review past transcriptions with original and AI-optimized text side by side. - **Custom Vocabulary** — Add professional terms, names, and product names. Auto-learning detects corrections and validates via LLM. - **Dark & Light Themes** — Dual theme support to match your preference. ## System Requirements - macOS 11.0+ - Apple Silicon processor recommended (for Metal GPU acceleration) ## Getting Started ### 1. Install Download the latest `.dmg` from [GitHub Releases](https://github.com/nicepkg/input0/releases), open it, and drag **Input 0** into your Applications folder. Using [Homebrew](https://brew.sh): ```shell brew install --cask input0 ``` > On first launch, macOS may show a security warning. Go to **System Settings → Privacy & Security** and click **Open Anyway**. ### 2. Grant Permissions Input 0 needs these macOS permissions to work properly. The app will prompt you on first launch: | Permission | Why | |---|---| | **Microphone** | Records your voice | | **Accessibility** | Simulates Cmd+V to paste text into other apps | Go to **System Settings → Privacy & Security** to grant each permission, then restart the app. ### 3. Configure LLM API Key ⚡ > **This step is required.** Without an API key, voice transcription still works, but the AI text optimization (grammar fix, filler removal, structuring) will be skipped. 1. Open Input 0 → Click the **⚙️ Settings** icon in the sidebar 2. Find the **LLM API** section 3. Enter your settings: | Field | Description | Default | |---|---|---| | **API Key** | Your OpenAI (or compatible) API key | *(empty — must fill in)* | | **API Base URL** | Service endpoint | `https://api.openai.com/v1` | | **Model** | LLM model name | `gpt-4o-mini` | **Using a third-party provider?** (e.g. Azure OpenAI, Groq, local Ollama, or any OpenAI-compatible API) - Change the **API Base URL** to your provider's endpoint - Set the **Model** to your provider's model name - Enter the corresponding **API Key** Click **Test Connection** to verify your API key works. ### 4. Download an STT Model ⚡ > **This step is required.** You need at least one model to transcribe voice to text. 1. Open Input 0 → Go to **Settings** → **Model** section 2. The app recommends the best model based on your language setting — follow the recommendation or pick your own 3. Click **Download** n
Excerpt of 7,348 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:3016c91f9041e325, topic:stt, topic:whisper, topic:speech-to-text