updated 04:25 pm EDT, Tue October 11, 2011
Could be linked to replacing Nuance
The US Patent and Trademark Office has granted Apple a new patent for a text-to-speech conversion process. Titled Multi-unit approach to text-to-speech synthesis, the patent describes a way of matching units from an input string to an audio library. An important inclusion is metadata, such as articulation relationships, which can inform a processor how to make phrases sound more natural. The software should also support a client-server architecture, allowing remote processing.
The patent could indicate that Apple eventually wants to rely on its own text-to-speech technology, rather than use licensed code from Nuance. Nuance's work is believed to form a smaller portion of Siri, the voice command system present on the iPhone 4S. The Nuance technology may also be present in OS X Lion, which has significantly better synthesized speech than Snow Leopard.
Because Siri is modular, Apple could theoretically replace any Nuance components with its own once they're sufficiently developed. Apple has increasingly preferred to use its own hardware and software in recent years, even going to the extent of designing its own processors. This not only allows for custom-tailored products but makes them harder to copy, especially because Apple controls the ideas involved.