Research
Research focus areas
The problems we study to improve voice, video, inference and multimodal AI.
Real-time speech intelligence
Streaming recognition and conversational responsiveness.
Multilingual models
Speech and language systems that work across languages.
Expressive synthesis
Natural, controllable speech generation.
Video understanding and generation
Analyzing, searching and creating visual content.
Efficient inference
Serving models with less compute and lower cost.
Multimodal interactions
Combining voice, vision and text in one experience.
Model safety
Reducing misuse and harmful outputs.
Compute optimization
Making better use of accelerators and infrastructure.
Publications and collaboration
We share papers, datasets and benchmarks here when results are ready to stand up to scrutiny. We welcome collaboration with universities and research labs working on these problems.
Collaborate with Aqylon Inc.
Interested in our research areas? We'd be glad to hear from you.