- Open, affordable AI air traffic control for flight simulators. Built in public with self-hosted and hosted - paths on the roadmap. + Push-to-talk in your sim, get authentic instructions back. Speech-to-text, an AI-guided state machine and + realistic text-to-speech keep every exchange inside real-world procedures.
- Collaborators welcome (Node/Nuxt, ATC SMEs, testers, infra benchmarking). Say hi at + We build in the open and welcome collaborators – engineers, ATC SMEs, voice nerds and sim pilots. Say hi at info@opensquawk.de.
- Open-source, low-cost AI ATC for flight simulation. Community-driven features, transparent costs and the - freedom to self-host or use hosted options later. + OpenSquawk combines open tooling with domain rules so every call feels like talking to a human controller – + without giving the LLM free rein.
- We are building an affordable core for radio analysis, decision logic and voices that anyone can inspect - and extend. + Whisper-class STT cleans up radio audio, the LLM matches it to our curated decision tree and Coqui/Piper + voices read back the exact instruction.
- Roadmap votes and community proposals steer our priorities – training, tools and integrations grow with - you. + Our MSFS plug-in mirrors heading, altitude, flight plan and traffic data into the controller so ATC can + vector, warn or hand off even without a pilot call.
- Beta goal: spin up your own instance via Docker within minutes or pick fairly priced hosting plans – open, - transparent, affordable. + Everything ships as open source so simmers, instructors and researchers can inspect prompts, rules and + voices – or self-host the full stack.
- We focus on Microsoft Flight Simulator first, work on the X-Plane concept in parallel and gather feedback - for additional platforms. + Live ATC launches with Microsoft Flight Simulator on Windows. We are mapping the same telemetry bridge to + MSFS 2024, then X-Plane and later FlightGear (Win/Linux/macOS).
- The current build runs locally (Node/Nuxt + services). You will need basic terminal skills and a taste for - experimentation. + The Classroom mode already runs listening drills: play ATC clips, extract clearances, compare with the + answer key and repeat until it sticks.
- Together with the community we structure training modules – from the first radio call to networks like - VATSIM. Feedback from tests flows straight into every chapter. + Invitation-only cohorts practise listening comprehension: extract the clearance, compare with the answer + key and repeat until every call sticks.
- Three ways to use OpenSquawk – all currently €0 while we cover the infrastructure. + Two experiences, one stack. Classroom is flying today via invitation, Live ATC is the flagship in active + development and the open-source core stays free to inspect or host yourself.
- Full control over infrastructure and data. Self-hosting is always an option. + AI controller that listens to your PTT call, consults the state machine and answers with authentic + phraseology. Perfect for solo IFR/VFR practice when networks are offline.
+ Listening drills that sharpen comprehension: decode clearances, practise readbacks and compare against + official transcripts. +
++ Clone the repo, run Docker and modify the controller logic. Ideal for researchers, virtual airlines or + devs who want a fully local stack. +
+- We cover the base costs so you can train with all fundamental features. For live ATC you provide your - own OpenAI API key – we only use it for the background radio requests. -
-- For virtual airlines and groups that want to practise together. We are testing dedicated environments - – currently also €0 and powered by your own API key. -
-- Direct VATSIM integration is on hold until licensing is clarified. We stay aligned with the network - guidelines before enabling any live connectivity. + Live ATC alpha runs offline-first while we research licensing for networks such as VATSIM and IVAO. We will + only enable direct connectivity once community rules and legal reviews are satisfied.
- In the meantime we focus on offline training flows and roadmap voting to decide which learning content - launches next. + Until then we double down on the state machine, simulator plug-ins and new Classroom scenarios – shaped by + roadmap votes and alpha feedback.
- Audio in → understanding → response out. Designed for low latency and clean radio discipline. + Every transmission passes through deterministic checkpoints so phraseology stays believable and safe.
Streaming speech-to-text tuned for radio.
Push-to-talk audio plus simulator telemetry stream into the engine.
LLM understands intent, context and state.
Whisper-class STT normalises radio noise; the LLM interprets intent.
Rules, flight data, taxi routing, validation.
A curated state machine enforces valid clearances and cross-checks flight data.
Natural voice-out with accurate numbers.
Coqui/Piper TTS deliver regionalised voices with crisp numbers.
- Current target is ~€4–5 per month, final pricing after cost benchmarking. We will share the results - transparently on the blog. + The Classroom is invitation-only during alpha. Join the waitlist and mention “Classroom” or email + info@opensquawk.de to request an invite.
- Open-source, low-cost AI ATC for flight simulators. Alpha prototype available – basic coding skills - recommended. + OpenSquawk builds Live ATC and Classroom training as open-source projects: speech-to-text, LLM-guided + state machine and natural TTS ready for sim pilots.