A self-adaptive intelligent voice device and method, wherein the device includes a voice recognizing module (101), a recognition result processing module (102), a voice broadcasting module (103), and a broadcast voice parameter generating module (104), and the broadcast voice parameter generating module (104) is configured to acquire extracted voice parameters from the voice recognizing module (101), and generate broadcast voice parameters according to the extracted voice parameters and a default policy, and input the broadcast voice parameters to the voice broadcasting module (103), the abovementioned technical solution enables the broadcast voice more closely matching a user's voice.