Mi Speaks

Independent developer Blaine Miller has launched an iPhone application named Mi Speaks, which converts text, digital books, and scanned images into spoken audio. The software stands out by running an 82-million-parameter artificial intelligence model directly on the handset rather than routing data to external cloud servers.

This local approach provides noticeable benefits for privacy and convenience. Users can convert sensitive work documents or personal reading material into speech while offline, ensuring that no private text leaves the device. Basic reading functionality is available for free using standard system voices, while a paid upgrade unlocks 28 neural voices, Apple Watch playback, and audio export capabilities.

It remains uncertain how heavily this local AI processing impacts iPhone battery life during lengthy listening sessions, especially when pre-rendering full documents. Additionally, as mobile processors become increasingly capable of running sophisticated neural models, it is unclear whether standalone apps can maintain a paid model if smartphone manufacturers decide to build equivalent private voice synthesis directly into future operating system updates.