The Scalable Language API enables conversant interaction using true natural language including speech recognition and speech synthesis. Sophisticated products incorporating the Scalable Language API will allow users to converse with products in their native language. SLAPI is both an API of defined linguistic functions and an architecture for processing language. SLAPI provides a consistent, well-documented architecture which is portable across platforms and broad enough to meaningfully cover every aspect of linguistic processing.
SLAPI's scope includes:
- Speech Recognition
- Speech Synthesis
- Syntactic Parsing
- Script Processing
- Notation for Writing Grammar (Lingua)
- XML & binary formats for lexicons & grammars
- Semantic Notation (InterLing)
- Logical Inferencing
- Test Corpora
- Runtime Metrics
- And much more.
The goal in creating the API was to define a standard API that is simple enough, and compact enough to be implemented in embedded products, yet broad enough and sophisticated enough to handle the full complexity of natural language, whether in a handheld, or on the desktop, or a large network server. The API can scale from a server down to a mobile phone.
Unlike other APIs which addressed only speech recognition and synthesis, the Scalable Language API encompasses all of natural-language processing. From the recognition and transcription of the speech, to the grammatical analysis, all the way through to language understanding, the SLAPI architecture covers the spectrum of requirements. In the synthesis direction, grammatical sentences can be composed and spoken. The engines provide the complex linguistic rules for the language grammar. The developer is freed from many of the complexities of natural language, and instead can concentrate on the semantics of the problem using high-level rules. For the first time, there is a commercial standard to allow developers to rapidly develop and test applications which can actually interact with the user through a Conversant User Interface (CUI) in their own language using ordinary sentences and vocabulary. This paves the way for some fascinating products using the new generation of processors which are so aptly suited for language processing.