Research Scientist, Addressee Detection​/Speech Separation

Research Scientist, Addressee Detection / Speech Separation

You will push the frontier on multi-speaker audio: separating overlapping voices and determining who is talking to the machine versus to another person in the room. Your work becomes the core of a product, not a paper that sits on a shelf.

• Advance our approach to addressee detection and speech separation in crowded, overlapping, real world audio.
• Frame the hard problems clearly, run focused experiments, and turn the results into models we can ship.
• Build the datasets and benchmarks that let us tell real progress from noise.
• Partner closely with engineering so that what works in research survives contact with production.

• Depth in speech, audio, or a closely related area of machine learning, whether from research, industry, or both.
• A track record of taking hard modeling problems from idea to result.
• Comfort working with real, messy audio rather than only clean benchmarks.
• Clear thinking and clear writing; you can explain a tradeoff to an engineer and to a customer.
• Interest in seeing your research reach real users quickly, at a small company.
• Bonus: published or shipped work in source separation, robust speech, or acoustic scene analysis.

attention labs is early: a small team defining a new category at the intersection of speech, cognitive neuroscience, and machine learning. We work from San Francisco, Toronto, and Memphis, and we are remote-friendly for the right person.