FIG. 4 is a figure for describing a case of recording a user's speech by the plurality of devices 20. In FIG. 4, it is supposed that an Internet TV 20A, a Home hub 20B, a desktop PC 20C, and a Laptop PC 20D are disposed as the plurality of devices 20 in Mr. A's room. Mr. A speaks the predetermined verbal start command (for example, Hello, °°, and then he speaks (asks) “How is the weather today?”, for example.

With respect to the Internet TV 20A, the Internet TV 20A is close to Mr. A and the surrounding noise is also small but a speaker is in use and there is an echo, for example, and therefore Reliability=Recording level (4)+Noise (5)+Echo (0)=9 is given.

With respect to the Home hub 20B, the Home hub 20B is close to Mr. A and is free from an echo and the surrounding noise has an intermediate level, for example, and therefore Reliability=Recording level (5)+Noise (3)+Echo (5)=13 is given.

With respect to the desktop PC 20C, the recording level is an intermediate level and a speaker is not used but the surroundings are noisy, for example, and therefore Reliability=Recording level (3)+Noise (1)+Echo (5)=9 is given.

With respect to the Laptop PC 20D, the Laptop PC 20D is free from the influence of echoes and the surrounding noise is also low but the Laptop PC 20D is distant from Mr. A and the recording volume is small, for example, and therefore Reliability=Recording level (2)+Noise (4)+Echo (5)=11 is given.

In this example, the reliability becomes higher in the order of Home hub 20B (Reliability=13)>Laptop PC 20D (Reliability=11)>Internet TV 20 A (Reliability=9)=Desktop PC 20C (Reliability=9).