How Human Can AI Sound? Try Sesame AI’s Conversational Speech Model Live
Sesame AI has made a significant leap in conversational voice technology, introducing a system that aims to cross the "uncanny valley" of AI-human interaction. On February 27, 2025, the company revealed its Conversational Speech Model (CSM), a breakthrough that promises to revolutionize how we interact with AI assistants.
Experience Sesame AI Voice Technology
Revolutionizing human-AI interaction with breakthrough technology that crosses the "uncanny valley" of conversation with emotional intelligence, natural timing, and authentic personality.
Revolutionary Features
Voice Technology Comparison
The Quest for "Voice Presence"
At the heart of Sesame's innovation is the concept of "voice presence" – the ability of AI to engage in genuine dialogue that builds trust and understanding over time. The CSM achieves this through:
- Emotional intelligence: Adapting to the emotional context of conversations
- Conversational dynamics: Incorporating natural timing, pauses, and emphasis
- Contextual awareness: Adjusting tone and style to match the situation
- Consistent personality: Maintaining a coherent and appropriate presence
Try to Talk to Sesame AI Now!
Conversational Voice Technology Demo
Experience the future of natural voice interaction, where conversations feel genuinely human. Press the call button below to start a conversation with either Maya or Miles, Sesame's advanced AI voice assistants.
The above demo is embedded directly from Sesame.com and is not hosted on this website
Microphone permission required. Calls are recorded for quality improvement but not used for ML training.
This demo is provided by Sesame AI. To experience the full demo in its original context, visit Sesame's website directly.
Technical Innovations
The CSM operates as a single-stage, multimodal learning system that combines text and audio processing. Key features include:
- Low-latency generation (200ms) for real-time interactions
- Pronunciation correction and homograph disambiguation
- Use of semantic and acoustic tokens for high-fidelity audio reconstruction
Public Demo and Reception
Sesame's research preview, featuring AI companions Maya and Miles, has garnered significant attention:
- The demo showcases human-like quirks, including filler words and contextual preferences
- Social media reactions have been overwhelmingly positive, with industry leaders praising the technology
- Journalists reported interactions so lifelike that bystanders mistook the AI for human conversation partners
Shopify CEO Tobi Lutke called the demo "absolutely insane," while Vercel CEO Guillermo Rauch described it as "astonishing."
Voice AI Comparison
Experience the difference between traditional AI voice systems and Sesame's revolutionary Conversational Speech Model (CSM)
Future Plans and Industry Impact
Sesame's ambitions extend beyond software:
- Development of AI eyewear for all-day wearable audio interaction
- Expansion of language support to over 20 languages
- Creation of duplex models for improved conversational flow
- Open-sourcing of key components under Apache 2.0 license
Founded by Oculus VR co-creator Brendan Iribe and speech technology expert Ankit Kumar, Sesame has secured Series A funding from Andreessen Horowitz and established offices in San Francisco, New York, and Bellevue.
Challenges and Ongoing Development
While the CSM has shown impressive results, Sesame acknowledges that challenges remain:
- Fully replicating human-like prosody in extended conversations
- Scaling up the model and dataset
- Developing truly duplex models for natural turn-taking
As the race for audio-first computing heats up, Sesame's innovations position it at the forefront of a potential paradigm shift in human-computer interaction. With its combination of technical prowess and visionary leadership, Sesame is poised to redefine our relationship with AI assistants, potentially making screen-based interfaces a thing of the past.
