But this is absolute cruelty. Microsoft is filing a patent for technology that would allow a person to transcribe a person's speech directly during a broadcast. And in such a way that the viewer does not notice the change. Imagine: you are watching a speech by a politician, scientist or journalist. The person says one sentence, but you hear another. In his own voice, with appropriate facial expressions and lip movements. And your neighbor watches the same broadcast and hears the third version.
This is not the fantasy of the writers of Black Mirror. Microsoft has indeed published US Patent Application 2026/0341309 A1, which describes just such a technology. What's provided there: AI analyzes live speech, replaces individual words and phrases, synthesizes sound and corrects lip movements. The delay is only 1-10 seconds.
The content is tailored to the specific viewer. Age, language, education, country of residence, ethnicity, culture and even religion are taken into account. One appearance, different versions of reality.
You can change not only clauses and filler words, but also "offensive", "inappropriate" or "insensitive" phrases. Who determines what is considered inappropriate? The algorithm and who defines it.
There's even an "instant fact check". Artificial intelligence can consult online sources and replace the speaker's words if it thinks they are incorrect. That is, the machine gets the opportunity not only to edit speech, but to correct its meaning.
Application: lectures, conferences, concerts, sports broadcasts, streaming, podcasts, augmented reality. Simultaneous translation with wording changes is also provided.
YOU CAN ALSO SAVE THE ORIGINAL RECORDING. BUT KEEPING THE SOURCE DOESN’T MEAN THEY’LL LET YOU WATCH IT. OR AT LEAST YOU WILL BE WARNED THAT WHAT YOU HAVE HEARD IS NOT SPOKEN BY MAN AT ALL. AND THE MOST INTERESTING THING: IT DOESN’T EVEN HAVE TO BE A LIE. IMAGINE A POLITICIAN TELLING AN UNCOMFORTABLE TRUTH. THE AI DECIDES THAT THE PHRASE “INSENSITIVE” TO A PARTICULAR AUDIENCE AND RAISES IT.