Audio data enhancement method and device, electronic equipment and storage medium

Abstract

The invention relates to an audio data enhancement method and device, electronic equipment and a storage medium, and the method comprises the steps: determining an audio recognition task which is a keyword detection task and / or a sound event detection task; receiving audio data associated with the audio recognition task; splitting and recombining the audio data according to the audio recognition task to obtain enhanced sample data for the audio recognition task; and obtaining an audio training sample for the audio recognition task according to the enhanced sample data and the audio recognition task. According to the invention, the audio data is split and recombined, and the obtained audio training sample has more prominent keyword features for a keyword detection task or more prominent sound features for a sound event detection task. The voice recognition accuracy of the keyword detection task can be improved, the detection response duration of the sound event detection task can be shortened, and the user experience of the keyword detection task and / or the sound event detection task can be improved.

Description

technical field [0001] The present disclosure relates to the field of computers, and in particular, to an audio data enhancement method, apparatus, electronic device, and storage medium. Background technique [0002] Currently, Key Word Spotting (KWS, Key Word Spotting) and Sound Event Detection (SED, Sound Event Detection) are two common speech tasks for edge smart speech devices. [0003] The keyword detection task needs to reduce the false wake-up rate while ensuring the detection rate. The sound event detection task requires as short a detection delay as possible, that is, the closer the detection time point is to the sound event occurrence time point, the better. [0004] Existing keyword detection tasks and / or sound event detection tasks are usually implemented using deep learning methods. Data augmentation is an important speech data processing method used in deep learning methods for keyword detection tasks and / or sound event detection tasks. Currently, data enhanc...

AI Technical Summary

Problems solved by technology

For example, for the event detection of a baby crying, the edge smart voice device and the baby are in one room, and the user is in another room for some reason, and needs to use the edge smart voice device to detect whether the baby is crying and enter the room where the baby is in time Take care of the baby. In this case, if the baby's crying sound is detected after a long time after the baby's crying, the user may not be able to learn the baby's crying in time and take corresponding measures in time

Images