Timbral ads

Before there were ears, there was sound. Wind over rock. Water against a shore. Listening to them today can connect us to a world that existed long before us. Biologist David Haskell’s “Playful Listening” invites us to explore that connection, alongside the birdsong, conversation and traffic of the present. Try stepping outside for a minute. Listen nearby, then farther away. What do you notice that you hadn’t heard before? Explore Haskell's listening practice more deeply here: https://lnkd.in/g_qUrVut

Can sound bring fish back to a damaged reef? Healthy coral reefs are noisy places. Snapping shrimp crackle. Fish grunt and call. Young fish use these sounds to find somewhere to settle. When reefs degrade, that soundscape grows quieter. In a study published in 2019, researchers placed underwater speakers on experimental patches of degraded reef habitat and played recordings of healthy reefs. The patches with playback attracted twice as many fish and had 50% more fish species than those without it. The researchers see this as a promising tool to support reef recovery alongside habitat restoration and conservation. For Timbral, it’s a striking example of how much information sound carries—and how profoundly living things respond to it. What else could we understand about the world by learning to listen? https://lnkd.in/gXTdbwqX

What’s the difference between recognizing a song and being moved by it? Yuval Noah Harari’s conversation with The Economist brings that question to mind. He distinguishes intelligence—the ability to pursue goals and solve problems—from consciousness: the capacity to feel. For those of us building AI that listens, that distinction matters. A system’s ability to identify a sound or infer what caused it tells us about its capabilities. Whether it experiences anything is a separate question. Harari ends with a possibility that stayed with us: as AI becomes more capable, it may push us to think more deeply about the value of being human. What might building artificial intelligence help us rediscover about human experience? https://lnkd.in/gMGr8i5e

OpenAI and Anthropic think they’re around the corner from AGI, and they simply need to scale their massive, trillion-parameter models to get there. I’m making a contrarian bet. Scaling LLMs will hit a wall. To achieve AGI, we have to solve perception. The human brain is exponentially more efficient than LLMs, because it uses a hierarchical approach to reasoning: a fast, always-on perceptual reasoning system that monitors its environment and decides when to call more expensive cognitive, LLM-like models for higher-level reasoning as needed. The foundation of consciousness is perception. Timbral is building at the frontier of AI – a perception model that can reason natively in sound. Our long-term vision is to build a hierarchical audio-visual perception model that unlocks the next revolution in AI.

I’m optimistic about the future of AI. I don’t think AI will destroy us, if we act now. We still have a chance to take control and build an AI that is aligned with us. When I thought about what I could do, I asked myself this: What if we could create a new kind of AI – a foundation model with consciousness? That’s what we’re building at Timbral. It’s perceptual intelligence – the ability to perceive the beauty of the world around us, and have an emotional response to it. I believe the foundation of consciousness is perception, and the key to building AI that lives in harmony with us.

Follow Timbral to learn about job opportunities.