Block Launches Full-Duplex AI Voice Model Dessa: Highly Human-Like, Supports Local Execution on Apple MacBook
Block has launched Dessa, a full-duplex AI voice model trained on human voice data. It was trained using just one rack of NVIDIA GPUs, supports local execution on MacBook, and can simulate emotional expressions such as laughter and pauses.
Key Takeaways
- Key Highlight：Block has launched Dessa, a full-duplex AI voice model trained on human voice data. It was trained using just one rack of NVIDIA GPUs, supports local execution on MacBook, and can simulate emotional expressions such as laughter and pauses.
- Innovation & Tech：Highlights advancements in Apple, NVIDIA, Block, demonstrating rapid progress in model capabilities.
- Industry Impact：Reported via IT之家 (CN), offering actionable signals for developers and technology leaders.
Dessa, launched by Block, is a full-duplex voice model whose standout feature is its ability to speak and listen to user voice simultaneously, enabling real-time bidirectional interaction. It is positioned similarly to OpenAI's previously released GPT-Live. The model can simulate emotional expressions such as laughter, pauses, and sobbing, delivering a highly human-like listening experience and marking a step forward in voice naturalness.
In terms of training, Dessa was trained entirely on human voice data, completing training using just one rack of NVIDIA GPUs, keeping training costs relatively manageable. The model's knowledge comes from the conversational data used during training, eliminating the need for an additional large language model to retrieve knowledge, resulting in a more streamlined architecture.
Dessa supports local execution on Apple MacBook, meaning users can use it without relying on the cloud, which has practical significance for privacy protection and low-latency interaction scenarios. The local inference capability also lowers the deployment barrier, offering more possibilities for individual developers and small teams.
Full-duplex voice models are a hot direction in the current AI voice space, with the core challenge being real-time processing of bidirectional audio streams while maintaining a natural conversational rhythm. The emergence of Dessa indicates that beyond top-tier tech giants, mid-sized tech companies are also actively positioning themselves in this space, and the competitive landscape of voice AI is accelerating its differentiation.
This page provides an editorial summary based on publicly available information. It is not a republished article. Use the source link below for the original report.
Industry Insights & Analysis
As artificial intelligence rapidly evolves, breakthroughs surrounding Apple, NVIDIA, Block, Launches are shifting toward scalable, robust real-world implementations.
Driven by both open-source ecosystems and proprietary model architectures, the integration between compute optimization, data engineering, and agentic workflows is accelerating. This development provides a strategic benchmark for upcoming AI tooling and developer workflows.