Overlapped voice and single voice distinguishing method based on long time characteristics and short time characteristics

Abstract

The invention discloses an overlapped voice and single voice distinguishing method based on long time characteristics and short time characteristics. The overlapped voice and single voice distinguishing method comprises the following steps: performing read-in of voice; performing voice pre-processing which comprises pre-emphasis, framing and windowing; extracting short time characteristic parameters, and extracting various short time characteristic parameters from each frame of voice; extracting long time characteristic parameters, and calculating statistical characteristics of the short time characteristic parameters; training a Gaussian mixture model; adopting an expectation-maximization algorithm to train four Gaussian mixture models; performing model fusion judgment, extracting the short time characteristic parameters and the long time characteristic parameters from tested voice to respectively serve as input of a short time characteristic model and a long time characteristic model, performing weighing of output probabilities of the two models to obtain a total probability output value, judging the tested voice to be the overlapped voice or the single voice according to the probability output value, and achieving distinguishing of the overlapped voice and the single voice. Compared with a method of adopting the short time characteristics, the overlapped voice and single voice distinguishing method has good distinguishing effect, and the distinguishing accuracy rate is improved by 5.9% on average.

Description

technical field [0001] The invention relates to speech signal processing and pattern recognition technology, in particular to a method for distinguishing overlapping speech from single-person speech based on long-term features and short-term features. Background technique [0002] Overlapped Speech (OS) refers to the speech produced when multiple people speak at the same time. Overlapping voices frequently appear in multi-person conversational voices. For example, in the ICSI conference voice database, 6-14% of the voices overlap. Due to the different acoustic characteristics of overlapping speech and single-person speech (speech produced by one speaker), the appearance of overlapping speech will lead to a sharp decline in the performance of current speech recognition systems and speaker segmentation and clustering systems that process single-person speech. Distinguishing overlapping speech from single-person speech is of great significance for improving the performance of ...

AI Technical Summary

Problems solved by technology

Although the feature parameters of the short-time frame layer can better describe some differences between overlapping voices and single-person voices, they cannot describe the differences in statistical significance between the above two voices, such as the mean difference, maximum value difference, and minimum value of features. value difference, median difference, mean square difference difference, etc.

In other words, short-term feature parameters cannot effectively represent the statistical difference between overlapping speech and single-person speech

Images