HaujetZhao/CapsWriter-Offline

PC 端语音输入工具，离线识别，高准确率、低延迟，支持热词、LLM润色。按住CapsLock或鼠标侧键X2说话，松开自动上屏。

What it solves

CapsWriter-Offline provides a completely offline, low-latency voice-to-text input system for Windows. It eliminates reliance on internet connectivity for speech recognition, ensuring privacy and accessibility on secure computers while providing a seamless "push-to-talk" experience for typing.

How it works

The project uses a client-server architecture where a server handles the speech recognition models and a client manages the keyboard input. Users hold a specific key (like CapsLock) to record audio, which is then processed by an offline engine. It supports multiple ASR (Automatic Speech Recognition) engines including Paraformer, SenseVoice-Small, Fun-ASR-Nano, and Qwen3-ASR, utilizing ONNX and GGUF formats for efficient local execution with optional GPU acceleration via DirectML or Vulkan.

Who it’s for

Windows 10/11 users who need fast, private, and highly customizable offline voice typing and transcription tools.

Highlights

  • Push-to-Talk Input: Instant text entry upon releasing the CapsLock key or mouse side buttons.
  • Offline Transcription: Ability to drag and drop audio/video files to generate .srt, .txt, or .json subtitles and transcripts.
  • Customizable Vocabulary: Hotword replacement using phoneme-based fuzzy matching and regex-based rules for precise terminology.
  • LLM Integration: Pre-configured roles (e.g., polishing or assistant) that process recognized text when specific trigger names are detected.
  • Digital ITN: Automatic conversion of spoken numbers into formatted digits (e.g., "fifteen or sixteen" to "15~16").
  • Privacy-First: All audio recordings are saved locally as files, ensuring no data leaves the machine.

Related

  • Project
    Gentleman-Programming/gentle-ai

    Gentle‑AI is a Go‑based configurator that enhances existing AI coding agents (Claude Code, Cursor, etc.) with persistent memory, skill libraries, planning (Spec‑Driven Development), security guardrails and optional evidence‑based review (Receipt‑Driven Development). It installs via a simple script, runs an interactive TUI to select agents and components, writes native config files, and provides a health‑check command. The tool does not ship its own model; it merely augments agents you already use.

  • Project
    akitaonrails/ai-memory

    ai‑memory is a Rust‑based server that records every observation a coding‑assistant makes (prompts, tool calls, session boundaries) as plain markdown files in a git‑backed wiki. The server can be run locally, in Docker/Podman, or on a homelab box and is designed to be shared across many agents (Claude Code, Codex, Cursor, Gemini CLI, etc.), machines and team members. Hooks posted by agents are consolidated into readable wiki pages at session end; the next session receives a concise handoff so it can continue exactly where the previous one stopped. Storage is human‑readable markdown plus a SQLite index (FTS5, entity and optional vector search). By default it needs no LLM or embedding keys; optional Anthropic/OpenAI integrations add LLM‑generated summaries and semantic search. Multi‑user authentication, audit logs and per‑person attribution are built in. Install via Arch AUR, Docker wrapper, or native binary, start the server, register a client and its hooks, and the system works transparently.

  • Project
    TalAter/annyang

    A tiny JavaScript library that enables users to control websites through voice commands using speech recognition.

  • Project
    espeak-ng/espeak-ng

    A compact, open-source text-to-speech synthesizer using formant synthesis to support over 100 languages with a very small memory footprint.