Block has introduced a new voice model designed to function locally on hardware, specifically optimized for the MacBook M4 Max. The company reported that the model is capable of generating voice output described as appearing “eerily human.”
By enabling the model to operate locally on the device rather than relying exclusively on cloud-based infrastructure, the development highlights a shift toward on-device artificial intelligence processing. This approach can offer potential benefits in latency reduction and data privacy, as audio processing remains contained within the user’s local hardware environment.
Technical specifications and the full scope of the model’s capabilities remain limited to the company’s disclosure regarding its local execution compatibility with the M4 Max architecture.
Source: TechCrunch
Original article: https://techcrunch.com/2026/10/24/blocks-new-voice-model-sounds-eerily-human/
Leave a Reply