Audio processing method and device, terminal and medium

Abstract

The embodiment of the invention discloses an audio processing method and device, a terminal and a medium, and relates to the technical field of audios. The audio processing method comprises the following steps: acquiring audio file data to be labeled; outputting a corresponding audio waveform on a display interface according to the audio file data; segmenting the audio waveform according to user operation to obtain at least two waveform areas; and determining a labeling result of the audio file data according to the label information corresponding to each waveform area. According to the embodiment of the invention, audio processing is more intuitive, simpler and more efficient.

Description

technical field [0001] The present invention relates to the field of audio technology, in particular to an audio processing method, device, terminal and medium. Background technique [0002] With the rapid development of information technology, existing information processing is not limited to simple media types such as text and pictures, and audio and video have also become one of the important sources of information processing. [0003] Specifically, in technical fields such as science and technology, machine learning speech recognition, etc., the processing of audio files is heavily involved. For example, in the field of machine learning speech recognition, the implementation of scenarios such as smart homes, smart devices, and smart customer service often requires training based on a large amount of audio annotation data, that is, the audio needs to be segmented and tagged. Among them, audio annotation is a technology that associates label information with specific audi...

AI Technical Summary

Problems solved by technology

[0004] The traditional audio annotation method can only rely on the annotator to play the audio file repeatedly, and then transcribe or classify the text based on the entire audio file content; while the audio segmentation lacks the support of actual tools, it can only be done by setting a predefined voice Parameters or pre-established rules, the audio is divided and output, which lacks flexibility, cannot be manually intervened, and does not meet the needs of data labeling

In order to meet the needs of data labeling, the existing effective alternatives are mainly to play the audio repeatedly, record multiple start and end timestamps and corresponding texts, but this method separates the audio and labeling results, and due to the low degree of visualization, When it is necessary to modify and adjust the marked area, it is necessary to find the corresponding area in all the recorded time periods first, and then make changes. The process is cumbersome and error-prone, which is not conducive to manual review, which greatly reduces the accuracy and accuracy of marking. Efficiency, difficult to cope with the growing demand for audio processing

Images