/ About this event
Voice AI is moving beyond synthetic narration and scripted assistants toward systems that can listen, reason, interrupt, adapt, and respond in real time.
But making a voice sound convincing is only one layer of the research problem.
How should we evaluate voice systems when naturalness, intelligibility, latency, expressiveness, speaker similarity, and task success pull in different directions? How do performance and preference change across languages, accents, domains, and noisy environments? What breaks when a model must handle turn-taking, interruptions, memory, grounding, and safety simultaneously?
This inaugural Voice Research Club brings together researchers and builders working across:
• Text-to-speech and speech recognition • Speech-to-speech and full-duplex systems • Voice agents and real-time interaction • Audio and multimodal foundation models • Evals, benchmarks, datasets, and human preference • Multilingual performance, safety, and enterprise deployment
The evening will feature two concise research presentations (second presenter to be announced) followed by an extended technical discussion. The talks provide the substrate; the Q&A is the main event.
This is a research forum. We will focus on methods, assumptions, evaluation design, results, failure modes, and what evidence would change our minds.
This event is a part of #SFTechWeek—a week of events hosted by VCs and startups to bring together the tech ecosystem. Learn more at www.tech-week.com.
