Voca - AI Dictation

The application converts spoken language to text directly in a web browser or via a native client for Windows, macOS, and Linux. Audio is captured locally when the user activates a hotkey, sent to a remote transcription service, and deleted immediately after the transcript is returned. Only minimal metadata—duration, timestamp, and model identifier—is retained for billing, and transcripts are stored only if the user opts in.

It targets users who need quick, on‑demand dictation while keeping their voice data private. The service offers a free quota of thirty minutes per month and a pay‑as‑you‑go model at ten cents per hour, with an unlimited subscription for six dollars a month. No background listening, backups, or training on user audio are performed.

The system emphasizes low latency, with median transcription times around 0.4 seconds, and operates without persistent logs or audio archives. The workflow consists of three steps: capture, transcribe, and return the text to the clipboard, providing a straightforward dictation experience.