Show Notes
AI voices can already sound startlingly human. The harder problem is whether they actually understand how humans speak.
Andrew Ettinger, CEO of Hume AI, joins Corey Noles and Grant Harvey to explain why voice AI still has a “listening problem.” A transcript can capture the words someone says while stripping away tone, hesitation, frustration, background noise, dialect, speaker identity, and the other signals that shape what those words actually mean.
The conversation explores how Hume evaluates voice systems against real-world conditions, why emotion is better understood as a high-dimensional spectrum than a handful of neat labels, what breaks during long-running voice interactions, how speech-to-speech systems and AI voice agents could become a primary interface for computing, and what new safety and personalization challenges appear when AI starts talking back.
If you want to understand what has to improve before voice AI feels truly natural—not just realistic—this episode is a practical look at the infrastructure, evaluation, and human feedback behind the next generation of spoken AI.
Subscribe to The Neuron at https://theneuron.ai for more AI news and analysis.
Sponsored by SAS: Grow faster and scale confidently with trusted AI solutions. Visit www.sas.com/smb
Sponsored by Origin: Origin is endpoint AI observability. See what a trace looks like at https://www.originhq.com/aiexplained
Sponsored by Outshift: “Scaling Out Superintelligence” Vijoy Pandey, January 2026. The technical whitepaper detailing the Internet of Cognition architecture, three-layer stack, and cognition state protocols. https://outshift.cisco.com/internet-of-cognition/whitepaper?utm_campaign=fy27q1_outshift_ww_paid_ioc-neuron-wp_podcast&utm_channel=podcast&utm_source=podcastInternet of Cognition Interactive Demo Clickable walkthrough showing per-agent activity, intent, context, and collective reasoning across a multi-agent SRE system.https://outshift.cisco.com/internet-of-cognition/explore?utm_campaign=fy27q1_outshift_ww_paid_ioc-neuron-explore_podcast&utm_channel=podcast&utm_source=podcast
More from The Neuron: AI Explained
Neurosurgeon-scientists Alex Ksendzovsky and Jon Pomeraniec explain how The Biological Computing Company uses living neurons as a research platform for AI. Their team grows roughly 150,000 cells on a 4,096-electrode array, sends in patterns such as handwritten digits, records how the neurons respond, and models useful computational behaviors in software. They say those neuron-inspired techniques are already helping GPU-based video-generation systems run about 5x faster with roughly 80% lower inference cost, while their longer-term work explores robots, world models, and real-time biological compute.
OpenAI just dropped 20+ announcements at DevDay 2026, so... yeah, we have some things to talk about. 😺
Runway co-founder Anastasis Germanidis explains Solaris, an “interface world model” that generates software frame by frame instead of rendering a traditional interface from HTML and CSS.
OpenAI DevDay 2026 is here, and we’re hosting a watch party to watch the keynote LIVE.
Corey and Grant sit down with Mammoth Biosciences co-founder and CEO Dr. Trevor Martin to explain how CRISPR is moving from a laboratory tool toward one-time medicines.