Can you help us understand how the Microsoft Speech Server handles special characters (-, \, <>, _, ‘etc.) and special words (acronyms [MCP, ER], roman numbers [I, II, III], etc.) are handle in the transition from the ARPA file to the compiled grammar?
For acronyms, when are the separated letter generated and when the whole word pronunciation is generated?
- To recognize the phrase “this is an abbreviation ABCT”, I must say each individual letters A B C T.
- To recognize the phrase “this is an abbreviation SIM”, I can’t say the letters S I M, I must say the word sim.
- To recognize the phrase “this is an abbreviation ABCI”, I can say the letters A B C I or the word abci.