There’s now AI technology that can recreate peoples faces using only voice recordings, and the results are eerily accurate.
It’s official. Computers know too much. Researchers at MIT’S Computer Science and Artificial Intelligence Laboratory (CSAIL) have created AI technology called Speech2Face that can guess what you look like based on your voice.
If the concept isn’t freaky enough already, wait until you see some of the results.
A demo of the software was originally published in a scientific paper written by CSAIL, allowing users to upload a snippet of audio which would then be converted into an AI portrait.
The technology has its obvious ethical issues, but CSAIL have defended those claims, stating that the AI “cannot recover the true identity of a person from their voice.”
“This is because our model is trained to capture visual features (related to age, gender, etc.) that are common to many individuals, and only in cases where there is strong enough evidence to connect those visual features with vocal/speech attributes in the data (see “voice-face correlations” below),” the company explained.
“As such, the model will only produce average-looking faces, with characteristic visual features that are correlated with the input speech. It will not produce images of specific individuals.”
The results still have a long way to go until they’re completely accurate, with common errors occurring such as misgendering the voice, mismatching ethnicity, and mistaking the subject’s age.
But some people are hypothesising that if improvements are made, the software could be a useful tool for investigators when solving crimes.
Try out Speech2Face for yourself, if you dare.
