| 摘要 |
PURPOSE: A method for loading a voice recognizing DSP program is provided to decrease a loading time and diagnose a memory easily through a check sum, by converting a COFF type executing program and a voice recognizing data portion composed by a different binary file into a binary file and loading the file including a check sum. CONSTITUTION: A voice recognizing DSP program composed by a COFF form executing program(T10) created through compiling of a source program and a text and binary type voice recognizing data(T11) are converted into a binary type executing program and voice recognizing data by passing a binary creating tool of a PC or a work station, and the voice recognizing data are transmitted to a storing unit of a voice recognizing system(T20, T30, T31). A check sum is included in a variated binary file, and when a recognizing board is mounted or re-driven, a main control unit which is an upper processor loads binary files stored in a storing device to a memory of a DSP(T40). If the binary files are loaded and performs a memory diagnosis command(T50), a performing process of the DSP reads information as a size of the binary data in the loaded binary file and compares to the check sum, and reports a diagnosis result of the memory to the main control unit(T60). If the voice recognizing DSP program is loaded normally, the voice recognizing DSP appoints a pointer of structure type variable, and allocates voice recognizing data value at once automatically, and performs an initialization and a voice recognizing operation(T70).
|
