Speech2Face: Learning the Face Behind a Voice

Publication/Creation Date
May 23 2019
Media Type
Journal Article
Persuasive Intent

How much can we infer about a person's looks from the way they speak? In this paper, we study the task of reconstructing a facial image of a person from a short audio recording of that person speaking. We design and train a deep neural network to perform this task using millions of natural Internet/YouTube videos of people speaking. During training, our model learns voice-face correlations that allow it to produce images that capture various physical attributes of the speakers such as age, gender and ethnicity. This is done in a self-supervised manner, by utilizing the natural co-occurrence of faces and speech in Internet videos, without the need to model attributes explicitly. We evaluate and numerically quantify how--and in what manner--our Speech2Face reconstructions, obtained directly from audio, resemble the true face images of the speakers.
HCI Platform
Discursive Type
Location on Body
Not On The Body
Marketing Keywords

Date archived
June 14 2019
Last edited
November 3 2019
How to cite this entry
Massachusetts Institute Of Technology (MIT), Computer Science and Artificial Intelligence Lab (CSAIL), Tae-Hyun Oh, Tali Dekel, Changil Kim, Inbar Mosseri, William T. Freeman, Michael Rubinstein, Wojciech Matusik. (May 23 2019). "Speech2Face: Learning the Face Behind a Voice". Computer Vision and Pattern Recognition. Fabric of Digital Life. https://fabricofdigitallife.com/index.php/Detail/objects/3927