SQS Signal — Self-Hosted Open Source Dictation App

SQS Signal

self-hosted dictation · open source · powered by Whisper

Speak your draft faster than you'd type it.

SQS Signal is an open-source dictation app that turns speech into text using OpenAI Whisper. Self-host it for full privacy and control — no cloud dependency, no data leakage.

open source self-hosted speech-to-text Whisper

Why dictation?

Speaking is often faster than typing — 2-3x faster for rough drafts, meeting notes, code comments, and brain-dumping ideas. SQS Signal gives you a simple, private way to do that on your own hardware.

Architecture

Three-container stack

Browser

→

Caddy
:80/:443

→

Web App (FastAPI)
:8000

→

Whisper Service
:8000

Caddy terminates TLS and reverse-proxies to the FastAPI web app. The web app handles auth, sessions, file uploads, and WebSocket live transcription, forwarding audio to a dedicated Whisper inference service.

Build: docker compose build web — the Python code is baked into the Docker image via COPY . ..

Mic mode — Record short clips (5-20s ideal) for quick CPU notes
Upload mode — Longer or higher-quality audio (up to 500 MB) using the better small.en model

Daily mic quota: 15 min per user; daily upload quota: 2000 MB
Language selection (11 languages + auto-detect)
Anti-hallucination filtering for short clips

Open-source stack: OpenAI Whisper · whisper.cpp compat · faster-whisper compat · FastAPI · Caddy

Support development:

☕ Buy me a Ko-fi