PromptVoice Local Voice Input
Hold a hotkey, speak, and get AI-polished text at your cursor
- Problem
- Users want a voice input utility not tied to an IME: hotkey to invoke, hold to speak, release to output—rather than some tools' tap-to-convert flow. Open-source options lack Chinese-focused tuning, FluidVoice mainly targets Western communities, Handy uses Rust but wraps a WebView, and WeChat's built-in voice input relies on the cloud and only works inside WeChat.
- Solution
- A local-first macOS voice input tool: hold a hotkey to record, transcribe via local or switchable ASR models, then polish with AI (punctuation, spoken-word cleanup, tone presets) and insert directly at the cursor—optimized for Chinese-first scenarios.
- Users
- Developers, writers, support and ops staff who type a lot in Chinese but don't want to switch IMEs, plus privacy-sensitive users.
- MVP
- Hold-hotkey recording and release-to-insert; local ASR model (switchable to cloud API); AI polishing toggle and tone presets; Chinese punctuation and spoken-word cleanup; menu bar residency and customizable hotkeys.
- Revenue
- One-time purchase ($9.99–$19.99) or free with premium AI polishing subscription; enterprise edition offers local model deployment packages.
- Why now
- Local ASR model quality is improving fast, making low-latency on-device voice input feasible; existing tools still have clear gaps in Chinese tuning and interaction gestures—ideal for an indie developer to own the niche experience.