An experiment successfully issued voice commands from the ground to a European Space Agency (ESA) satellite and launched the FPS game “DOOM” in space. The system worked by having AI onboard the satellite recognize the voice and launch the game. Seven experiments were conducted from April through July 2026, ultimately succeeding in control through speech spoken directly by a human.
First Voice Command of a Satellite Achieved Through ESA’s OPS-SAT Space Lab - Tanagra Space
https://tanagraspace.com/press/first-voice-command-of-a-satellite/
Oslo Amateur Radio Group achieves world’s first voice control of a satellite! – Oslogruppen av NRRL
https://la4o.no/oslo-amateur-radio-group-achieves-worlds-first-voice-control-of-a-satellite
PRETTY DOOMed: The First Voice Command of a Satellite - Georges Labrèche
https://georges.fyi/opssat/first-voice-command-of-a-satellite/?utm_source=chatgpt.com
Attempts to run “DOOM” on satellites had been underway for some time, and in 2023, the game was successfully executed on another satellite, “OPS-SAT-1.”
Running DOOM on a satellite - YouTube
This experiment took a step further, aiming to have the satellite recognize human voice transmitted from the ground and launch “DOOM” in accordance with those commands.
The ESA experimental satellite used in this experiment, “OPS-SAT PRETTY,” was originally intended for observations using radio-wave reflections from positioning satellites and for monitoring radiation, and it was not equipped with voice-control capabilities. However, because it had a computer whose software could be rewritten and an SDR (software-defined radio) that allowed radio-communications processing to be modified through software, new functionality could be added.
The experiment was led by Tanagra Space, which develops machine-learning technology for space, in cooperation with ESA, Graz University of Technology, and amateur radio operators. Between April and June 2026, an amateur radio station in Regnitsa, Poland, conducted four experiments. An amateur radio group in Oslo, Norway, “LA4O,” then took over the experiment.
On May 29, 2026, the Oslo team was approached to assist with the experiment and prepared radio equipment, amplifiers, antennas, and other gear. It set up a system that transmits voice at approximately 80 W of output using a frequency of 1296 MHz for communications from the ground to the satellite.
On July 3, in the fifth experiment, the satellite was able to receive radio signals from the ground, but there was a lot of noise in the recorded audio, making it impossible to hear the contents. The cause was that the bandwidth used to process the radio signals was too wide for the audio signal, so the team devised a method to extract only the necessary frequency band and demodulate it. They submitted the corrected software to the ESA about a week later.
With the corrected software, audio was extracted from the radio signals received by the SDR, and transcription was performed using an approximately 27 MB AI speech recognition model. If “PLAY DOOM” was detected in the transcription, DOOM would launch automatically; thus, even if there were some errors in the recognition results, the command could still be detected.
In the sixth experiment, held on July 27, pre-synthesized speech was repeatedly transmitted instead of a human voice. As a result, the command was successfully detected in three of the six audio recordings made by the satellite, and DOOM was launched.
On the same night, the seventh experiment was conducted. This time, instead of audio recorded in advance, amateur radio operator John Bellgril Hyell spoke directly into a microphone: “LIMA ALFA FOUR OSCAR, PRETTY PLAY DOOM DOOM DOOM, PRETTY PLAY DOOM DOOM DOOM.” The opening phrase, “LIMA ALFA FOUR OSCAR,” was the radio station’s identification code, and the portion that followed was the command to launch DOOM.
The actual voice command spoken by Mr. Hyell can be heard below.
Voice command for launching DOOM on an artificial satellite in space - YouTube
The satellite received the audio and performed AI-based transcription. Although the transcription contained errors, it successfully recognized the word “DOOM” correctly twice. It launched DOOM as commanded, successfully demonstrating control of the satellite using a human voice.
The satellite generated and sent to the ground screenshots of the game screen, animations, audio transcripts, radio signal analysis data, and other data. It should be noted that DOOM operated by running a game demo preinstalled on the satellite; a human on the ground did not control the character in real time.
“The important thing in this experiment was not the voice itself, but the fact that the satellite could autonomously analyze the radio waves it received and take action,” explained Georges Labreuche of Tanagra Space. He said the same technology could be applied in the future to detecting distress signals sent by people in distress and identifying unusual radio signals coming from space, among other uses.
