Hey Product Hunt, I built SpeakCLI because typing `kubectl get pods --namespace prod` for the four-hundredth time felt insulting. Code assistants got smart fast, but the input layer is the same keyboard I used in 2003. At first i built this for myself until i realized: Let's publish this and make it a public project where the features and utility from the users matter most! From a Developer to Developers. The bet: voice is finally good enough to be a serious dev input — Whisper is accurate, fast, and runs on CPU. So I made the smallest thing that does one job well: press Ctrl+., speak, get text wherever your cursor is. No assistant, no interpretation, no cloud round-trip. Three things I sweated: 1. Latency — the full hotkey-to-text pipeline lands under a second 2. Privacy — audio literally never leaves your machine. No keys, no telemetry on speech 3. Pricing — Free, with a one-time $9.99 Pro Plan. (I'm tired of subscriptions) It pairs beautifully as the voice layer in front of Claude Code / Copilot CLI / Warp — you speak the intent, the AI does the work. Would love your feedback, especially edge cases I haven't hit yet. AMA in the thread.

SpeakCLI

Speak. It types. Local push-to-talk dictation for devs.

Speak. It types. Local push-to-talk dictation for devs.