Method and device for detecting voice quality

Abstract

One embodiment of the invention discloses a speech quality detecting method. The method comprises the following steps: receiving test task information sent by a calling terminal, wherein the test task information includes the file identifier of a test playback file and a test called terminal ID; searching a pre-stored standard corpus for a standard speech file corresponding to the file identifier of the test playback file; and when receiving a recording file sent by a called terminal corresponding to the test called terminal ID, performing contrast calculation on the standard speech file and the recording file to get a value of speech quality. The other embodiment of the invention discloses a speech quality detecting device.

Description

technical field [0001] The present invention relates to the field of voice services based on long-term evolution (VoLTE, Voice over Long Term Evolution), in particular to a voice quality detection method and device. Background technique [0002] With the advent of the VoLTE era of the 4th Generation mobile communication technology (4G), broadband voice quality issues have attracted much attention. For how to evaluate voice quality, the industry generally adopts the traditional Mean Opinion Score (Mean Opinion Score) , MOS) box test mode, the MOS box is the core equipment in this test scheme, and the MOS box is connected with a personal computer (Personal Computer, PC) and a mobile phone respectively. The PC controls the MOS box to play the playback file, and the playback file is transmitted to the calling mobile phone through the audio cable. The calling mobile phone undergoes signal processing, and transmits the processed voice signal to the called mobile phone through netw...

AI Technical Summary

Problems solved by technology

[0004] 1. The existing voice quality test is based on the MOS box equipment. Before the test, the MOS box needs to be connected to the PC and the mobile phone. That is, the engineer needs to carry two mobile phones, a computer, a MOS box and several connecting lines for the test. Lines are messy, which brings great inconvenience to user testing and low testing efficiency

[0005] 2. Since the MOS algorithm (such as the POLQA algorithm) is bound to the hardware device (MOS box), only the MOS box can be used for testing, and the MOS scoring can only be performed in the MOS box

Therefore, it is necessary to carry out MOS test or MOS scoring only with hardware equipment, which lacks flexibility.

[0006] 3. The caller and the caller need to be connected to the MOS box at the same time. The test scene is single, and the test results can only be used as a reference value for the voice quality of a certain period of time in a certain area. In reality, the caller and the caller are most of the time when the user is talking. Under the same coverage cell, the voice quality test across cells is particularly important and has practical application value, but this test solution cannot reproduce the problem of low voice quality when the caller and called party make calls at different locations

[0007] 4. The test results of the MOS box test method are only saved locally, that is, the data of each test engineer is basically known only to himself, and it is impossible to form a unified MOS voice test database to conduct multi-dimensional large-scale analysis of user terminal types, wireless scenarios, etc. Data analysis, making the application value of MOS test limited

[0008] 5. Even if the voice quality is poor by using the MOS box test method, it is still impossible to locate which section of the network has a problem that causes the overall poor voice quality, so it can only be used as a means of voice quality assessment, not as a voice quality Poor detection method, that is, you can only know where the voice quality is poor, but you can't know what caused it

Images