Techniques for determining direction of arrival for an emergency situation are detailed. A plurality of audio sensors of a vehicle may receive audio data associated with the vehicle. An audio sensor pair may be selected from a plurality of audio sensors to generate audio data representing sound in an environment in which the vehicle is located. Based on the audio data, an angular spectrum associated with the audio sensor pairing may be determined. Features associated with the audio data may be determined based on the angular spectrum and/or the audio data itself. Based on the feature, using a machine learning model, a direction of arrival (DoA) value associated with the sound may be determined. An emergency sound (e.g., an alarm sound) may be detected in the audio data, and a direction relative to the vehicle associated with the emergency situation may be determined based on the feature and the DoA value.
详细描述了用于确定紧急情况的波达方向的技术。车辆的多个音频传感器可以接收与车辆相关联的音频数据。可以从多个音频传感器中选择音频传感器配对以生成表示在车辆所处环境中的声音的音频数据。基于音频数据,可以确定与音频传感器配对相关联的角谱。基于角谱和/或音频数据本身,可以确定与音频数据相关联的特征。基于该特征使用机器学习模型,可以确定与声音相关联的波达方向(DoA)值。在音频数据中可以检测到紧急声音(例如警报声),并且基于该特征和DoA值,可以确定与紧急情况相关联的相对于车辆的方向。
Emergency vehicle detection
应急车辆检测
2022-06-03
Patent
Electronic Resource
Chinese