To provide an update data generation method of a speech recognition dictionary, for easily updating the speech recognition dictionary with a small update data amount.

A dictionary (n) of reading and a key data (n) are generated from a speech recognition original data (n-1) which has been used when generating a (n-1)th update data of previous time, and a speech recognition original data (n) which is used for generation of update data this time ( n-th) to make update data (n), and it is supplied to a navigation system via a suitable communication medium or a storage medium. In the key data (n), for a place added in the speech recognition original data (n), place information of the place is registered in the key data (n) for each place. In the dictionary (n) of reading, for all places, reading of a name of the place, and a key pointer for indicating entry of the key data which is generated this time or past time, when the place information of the place is registered, are registered.