Overview

Sonocles is a menu bar app and a small HTTP API. Launch it and two sockets come up on loopback — server-sent events on 7357, WebSocket on 7358 — and stay up for as long as the app runs. Start listening, from the popover or with POST /start, and every word you say arrives on both as a JSON frame carrying the text, a sequence number, the audio time it was spoken, and how far behind you it turned up.

It exists because Pteroprompter needed a speech-follow that did not fight the encoder for CPU. Nothing about it is prompter-shaped, though: it streams words and timestamps at a socket, and what listens is your business.

The shortest possible consumer

new EventSource('http://127.0.0.1:7357/events').onmessage = (e) => console.log(JSON.parse(e.data).text)

{ "type": "partial", "text": "the menu bar", "ts": 1788569799791, "seq": 3,
  "audioStart": 40.28, "audioEnd": 40.8, "lagMs": 200 }

Use audioEnd, not ts. A missing lagMs means unmeasured — never zero. We are quite firm about this, and the protocol page says why.

Where to go next

The pages in the sidebar are in reading order. The two that make Sonocles different from a dictation app are The protocol — what a frame carries and what each field is for — and Engines and the measurement — why it is Parakeet on the Neural Engine and not Apple’s Speech framework, with the numbers.