Man-machine interaction method and device

Abstract

The invention discloses a human-computer interaction method and device, and the method comprises the steps: collecting a first continuous image frame based on camera equipment; recognizing a face attribute by using the first continuous image frame, recognizing a distance value between the face attribute and the camera equipment, and judging whether the distance value is within an effective distance or not; when it is judged that the distance value is within the effective distance, the face orientation angle is determined through the face attributes, whether the face orientation angle is an effective angle or not is judged, and whether the time when the face orientation angle is the effective angle lasts for preset duration or not is judged; if the face orientation angle is the effective angle and the time when the face orientation angle is the effective angle lasts for a preset duration, collecting audio and judging whether the audio contains human voice or not, collecting a second continuous image frame at the same time, obtaining face features in the second continuous image frame, and judging whether the speaker is speaking or not at least based on the face features; and if the speaker is speaking, combining the audio and the facial features to obtain an effective audio.

Description

technical field [0001] The invention belongs to the field of artificial intelligence, and in particular relates to a human-computer interaction method and device. Background technique [0002] Human-computer voice interaction in public places (such as subway ticket machines, vending machines, large screens in exhibition halls, etc.), there is a lot of background noise. At present, there are methods of rejecting long-distance noise based on front-end signal processing technology of microphone array and methods of rejecting noise based on lip movement. [0003] Among them, based on the front-end signal processing technology of the microphone array, the microphone array hardware device is configured to reject long-distance noise, and the audio is collected. Through signal processing technology, the position of the sound source is calculated, and the audio of the far sound source is rejected according to the position of the sound source, so that Achieve noise rejection. If the...

AI Technical Summary

Problems solved by technology

If the speaker is speaking within a relatively short distance, the technology cannot distinguish whether the speaker is speaking to the interactive device, cannot determine whether the speaker's focus is on the interactive device, and cannot distinguish which are the invalid audio of the speaker

Images