Natural language interfaces

A speech processing system, comprising:

• one or more modular subsystems comprising one or more processors, the modular subsystems comprising,
• an automatic speech recognition module configured to receive audio input with speech and further configured to output text;
• a domain-specific lexicon defining a domain-specific vocabulary;
• a domain-specific grammar specification defining rules for using the domain-specific vocabulary consistently with a domain agnostic language grammar specification;
• a grammar projection module configured to:
• receive the text,
• generate corrected text by projecting the text against the domain-specific grammar specification, with the projecting including;
• generating multiple candidate sentences by applying the text to the domain-specific grammar specification,
• determining a phonetic distance between the text and one or more of the multiple candidate sentences,
• selecting a particular sentence of the multiple candidate sentences based on the phonetic distance, and
• configuring the particular sentence as the corrected text, and
• output the corrected text;
• and
• an interpreter module configured to receive the corrected text and evaluate the corrected text.