Teaching computers to pick a conversation out of a noisy room

Human beings have a weird, poorly-understood ability to pick a single conversation out of a noisy room, it's called the "cocktail party effect" and while its exact mechanism isn't totally understood, researchers do know that vision plays a role in it, and that being able to see the speaker helps you pick their words out of a crowd.

So a group of computer scientists decided to replicate the feat: using a machine-learning system, they created an interface that lets users choose a person in a video in which several people are speaking, and then, using a lip-reading algorithm and other visual cues, pulls the individual audio track out of the mass, isolating it.

We present a joint audio-visual model for isolating a single speech signal from a mixture of sounds such as other speakers and background noise. Solving this task using only audio as input is extremely challenging and does not provide an association of the separated speech signals with speakers in the video. In this paper, we present a deep network-based model that incorporates both visual and auditory signals to solve this task. The visual features are used to “focus” the audio on desired speakers in a scene and to improve the speech separation quality. To train our joint audio-visual model, we introduce AVSpeech, a new dataset comprised of thousands of hours of video segments from the Web. We demonstrate the applicability of our method to classic speech separation tasks, as well as real-world scenarios involving heated interviews, noisy bars, and screaming children, only requiring the user to specify the face of the person in the video whose speech they want to isolate. Our method shows clear advantage over state-of-the-art audio-only speech separation in cases of mixed speech. In addition, our model, which is speaker-independent (trained once, applicable to any speaker), produces better results than recent audio-visual speech separation methods that are speaker-dependent (require training a separate model for each speaker of interest).

Looking to Listen at the Cocktail Party: A Speaker-Independent Audio-Visual Model for Speech Separation [Ariel Ephrat, Inbar Mosseri, Oran Lang, Tali Dekel, Kevin Wilson, Avinatan Hassidim, William T. Freeman and Michael Rubinstein/Arxiv]

(via 4 Short Links)