Intelligent vehicle ADAS early warning test method and system based on sound recognition

Abstract

The invention provides an intelligent vehicle ADAS early warning test method and system based on sound recognition. The method comprises the steps of S1, acquiring an alarm sound, wherein the alarm sound comprises K alarm sub-sounds; S2, carrying out sound extraction processing on the k alarm sub-sound obtained in the step S1, and forming an alarm sound database by the extracted sound data; S3, acquiring to-be-processed sound, and preprocessing the acquired to-be-processed sound to obtain pre-processed sound; S4, performing sound extraction processing on the preprocessed sound obtained inthe step S3 to obtain to-be-compared sound data; and S5, comparing the to-be-compared sound data obtained in the step S4 with the alarm sound database in the step S2, and if the comparison similarityis greater than or equal to preset similarity, determining that the to-be-processed sound is an alarm sound. By means of the early warning test method, early warning sounds can be extracted, a drivercan take measures early, and safe driving of the driver is achieved.

Description

Technical field [0001] The present invention relates to the technical field of intelligent vehicles, in particular to an ADAS early warning test method and system for intelligent vehicles based on voice recognition. Background technique [0002] Smart cars' ADAS (Advanced Driving Assistant System) functions are becoming more and more abundant. In the initial stage of the development of smart cars, the ADAS function is mainly used to remind the driver through voice warning, so that the driver can take early measures to avoid danger happened. [0003] In the current test, we first estimate the frequency range of the alarm sound through the sound frequency recognition software, and then set the frequency range in the sound collection test device. When the driving vehicle triggers the alarm sound, the sound collection device performs analog-to-digital conversion. It appears in the form of waves. In the later data processing, the alarm moment in the square wave is identified through ex...

AI Technical Summary

Problems solved by technology

There are three problems in the above sound alarm time recognition method: first, according to the sound frequency recognition software can only estimate the frequency range of the alarm sound, the accuracy is low, and the testers often have to try and adjust repeatedly to achieve better results; second, After the sound collection equipment is converted into a square wave, the waveform characteristics at the alarm time may not be obvious, and there may be interference caused by other glitches or background noise.

Unobvious alarm time characteristics are not conducive to the automatic identification of alarm time by the post-processing program, which requires a lot of manual labor to find the alarm time based on experience

Third, this sound collection method can only collect one sound frequency in one test after setting the frequency range. If it is necessary to know the sounding time of two or more sounds at the same time in the same collection, is not allowed

Images