WO2017041553A1 - Method and apparatus for determining voice quality - Google Patents

Abstract

A method for determining voice quality and an apparatus for determining voice quality. The method comprises: acquiring, during voice quality measurement, a voice correlation parameter of a test point in the measurement (101); and determining voice quality of the test point according to the acquired voice correlation parameter and a preset evaluation parameter influencing a voice (102). The efficiency of a voice quality test can be improved, and the measurement cost of the voice quality test is reduced.

Description

Method and device for determining voice quality Technical field

This document relates to, but is not limited to, audio communication technology, and more particularly to a method and apparatus for voice quality determination.

Background technique

As the basic component of mobile communication, voice communication has always been a hot spot of attention. Therefore, researchers have proposed many methods for measuring voice quality in voice communications.

In the related art, the corpus is generally sent from the call origin by using a voice quality test instrument, and the call end simulates the subjective listening experience of the human ear for evaluation.

The above method of using the subjective listening experience for voice quality evaluation has a large test workload, and the voice quality test instrument has a high price, that is, the voice quality test has low efficiency and high cost.

Summary of the invention

The following is an overview of the topics detailed in this document. This Summary is not intended to limit the scope of the claims.

The embodiments of the present invention provide a method and an apparatus for determining voice quality, which can improve the testing efficiency of the voice quality test and reduce the testing cost of the voice quality test.

In a first aspect, an embodiment of the present invention provides a method for determining a voice quality, where the method includes:

When performing voice quality measurement, obtain the voice correlation parameter of the test point in the current measurement;

Determining the voice quality of the test point according to the obtained voice correlation parameter and the preset evaluation parameter affecting the voice.

Optionally, the preset evaluation parameters affecting the voice include a voice evaluation parameter and an environment correction parameter;

Determining, according to the obtained voice correlation parameter and the preset evaluation parameter that affects the voice, the voice quality of the test point includes:

And the obtained voice correlation parameter, the voice evaluation parameter, and the environment correction parameter Substituting a quality formula to obtain a speech quality of the test point, the quality formula is: M=A*B+c;

Wherein said M is a speech quality of said test point; said A is a 1×n matrix comprising n speech evaluation parameters; said B being an n×1 matrix comprising n speech correlation parameters; c is an environmental correction parameter, and n is a positive integer.

Optionally, the voice evaluation parameter includes an overlay correction coefficient a1, an interference correction coefficient a2, a transmission quality correction coefficient a3, and a handover delay correction coefficient a4;

The voice correlation parameter includes a reference signal received power b1, a signal to interference plus noise ratio b2, a block error rate b3, and a switching interval time b4, which is a time interval from the cut-in cell to the current measurement. ;

The corresponding mass formula is: M = a1 * b1 + a2 * b2 + a3 * b3 + a4 * b4 + c.

Optionally, before the acquiring the voice correlation parameter of the test point in the current measurement, the method further includes:

Performing multiple regression according to the voice correlation parameters of the two or more sets of the test points stored in advance, and the real voice quality corresponding to the voice correlation parameter of each set of the test points and the quality formula, and obtaining the prediction Set evaluation parameters.

Optionally, the method further includes:

After the voice correlation parameter, the voice evaluation parameter, and the environment correction parameter are substituted into the quality formula to obtain the voice quality of the test point,

Obtaining the true voice quality of the test point;

Determining whether a difference between the voice quality of the test point and the real voice quality is less than a preset threshold;

When the difference between the voice quality of the test point and the real voice quality is greater than or equal to a preset threshold, the voice correlation parameter of the test point and the real voice quality of the test point are saved as a set of data;

Performing multiple regression according to the voice correlation parameter of each set of the test points, the real voice quality corresponding to the voice correlation parameter of the test point, and the quality formula, to obtain a new evaluation parameter;

The preset evaluation parameter is updated with the obtained new evaluation parameter.

In a second aspect, the embodiment of the present invention further provides a device for determining a voice quality, where the device includes:

The obtaining unit is configured to obtain a voice correlation parameter of the test point in the current measurement when performing voice quality measurement;

The determining unit is configured to determine the voice quality of the test point according to the obtained voice relevance parameter and the preset evaluation parameter that affects the voice.

Optionally, the preset evaluation parameters affecting the voice include a voice evaluation parameter and an environment correction parameter, and the determining unit is configured to:

Substituting the voice correlation parameter, the voice evaluation parameter and the environment correction parameter into a quality formula to obtain a voice quality of the test point, the quality formula is: M=A*B+c;

Wherein said M is a speech quality of said test point; said A is a 1×n matrix comprising n speech evaluation parameters; said B is an n×1 matrix comprising n speech correlation parameters, c is an environmental correction parameter, and n is a positive integer.

Optionally, the voice evaluation parameter includes an overlay correction coefficient a1, an interference correction coefficient a2, a transmission quality correction coefficient a3, and a handover delay correction coefficient a4;

The voice correlation parameter includes a reference signal received power b1, a signal to interference plus noise ratio b2, a block error rate b3, and a switching interval time b4, which is a time interval from the cut-in cell to the current measurement. ;

The corresponding mass formula is: M = a1 * b1 + a2 * b2 + a3 * b3 + a4 * b4 + c.

Optionally, the device further includes:

The first calculating unit is configured to perform, according to the voice correlation parameter of the two or more sets of the test points stored in advance, and the real voice quality and the quality formula corresponding to the voice correlation parameter of each group of the test points. Multiple regression, the preset evaluation parameters are obtained.

Optionally, the acquiring unit is further configured to acquire a real voice quality of the test point;

The device also includes:

The determining unit is configured to determine whether a difference between the voice quality of the test point and the real voice quality is less than a preset threshold;

The saving unit is configured to: when the difference between the voice quality of the test point and the real voice quality is greater than or equal to a preset threshold, the voice correlation parameter of the test point and the real voice quality of the test point are taken as one Group data is saved;

The second calculating unit is configured to perform multiple regression according to the voice correlation parameter of each set of the test points, the real voice quality corresponding to the voice correlation parameter of the test point, and the quality formula, to obtain a new evaluation parameter;

The update unit is configured to update the preset evaluation parameter with the obtained new evaluation parameter.

Compared with the related art, the technical solution provided by the present invention includes: when performing a voice quality test, obtaining a voice correlation parameter of a test point in the current measurement; and then, according to the acquired voice correlation parameter and a preset evaluation parameter affecting the voice Determine the voice quality of the test point. In the embodiment of the present invention, in the process of performing voice quality testing, it is no longer necessary to use the measuring instrument to measure the test point, but the voice quality of the test point is estimated according to the voice correlation parameter of the test point and the preset evaluation parameter. Since the measuring instrument is no longer needed, the efficiency of the voice quality test is improved, and the measurement cost of the voice quality test is reduced.

Other aspects will be apparent upon reading and understanding the drawings and detailed description.

BRIEF abstract

FIG. 1 is a flowchart of a method for determining voice quality according to an embodiment of the present invention;

2 is a flowchart of another method for determining voice quality according to an embodiment of the present invention;

FIG. 3 is a schematic structural diagram of an apparatus for determining a voice quality according to an embodiment of the present invention;

FIG. 4 is a schematic structural diagram of another apparatus for determining voice quality according to an embodiment of the present invention.

Embodiments of the invention

Embodiments of the present application will be described in detail below with reference to the accompanying drawings. It should be noted that, in the case of no conflict, the features in the embodiments and the embodiments in the present application may be arbitrarily combined with each other.

Embodiment 1

The embodiment of the invention provides a method for determining voice quality. As shown in FIG. 1 , the method includes:

Step 101: When performing voice quality measurement, obtain a voice correlation parameter of the test point in the current measurement.

Here, the voice correlation parameter of the test point includes: parameters of the ongoing voice call at the test point; optionally, the voice correlation parameter may include: reference signal received power, signal to interference plus noise ratio, block error rate, and One or more of the switching interval time, etc.; wherein the switching interval is the time interval from the cut-in cell to the current measurement.

Step 102: Determine a voice quality of the test point according to the obtained voice correlation parameter and a preset evaluation parameter that affects the voice.

Optionally, when the preset evaluation parameter includes a voice evaluation parameter and an environment correction parameter, the voice correlation parameter, the voice evaluation parameter, and the environment correction parameter are substituted into the quality formula to obtain the voice quality of the test point, and the quality formula may be: M= A*B+c;

Where M is the speech quality of the test point; A is a 1×n matrix, including n speech evaluation parameters; B is an n×1 matrix, including n speech correlation parameters; c is an environmental correction parameter, and n is a positive integer .

Optionally, when the voice evaluation parameter includes a coverage correction coefficient a1, an interference correction coefficient a2, a transmission quality correction coefficient a3, and a handover delay correction coefficient a4; the voice correlation parameter includes a reference signal received power b1, a signal and an interference plus noise ratio b2 The block error rate b3 and the switch interval time b4; wherein the switch interval time b4 is the time interval from the cut-in cell to the current measurement; the corresponding quality formula is: M=a1*b1+a2*b2+a3*b3+ A4*b4+c, the quality formula here can only include some or all of them, and other items can be added. Here, the hand-in cell can cut in to the cell where the current terminal is located for the terminal that was last measured.

In this way, during each test, it is no longer necessary to use the measuring instrument to measure the test point, but according to the speech correlation parameter of the test point and the preset evaluation parameter, the voice quality of the test point can be estimated. Since the measuring instrument is no longer needed, the measurement cost is greatly reduced.

Optionally, the embodiment further needs to initialize the preset evaluation parameter. Therefore, before the step 101, the method of the embodiment of the present invention further includes: according to the voice correlation parameter of two or more sets of test points stored in advance, and each group The real speech quality and quality formula corresponding to the speech correlation parameter of the test point are subjected to multiple regression, and the preset evaluation parameters are obtained.

It should be noted that the real voice quality may include: the real voice quality obtained by subjective evaluation and the real voice quality obtained by using the voice quality measuring instrument; the real voice quality obtained by subjective evaluation includes: people pass the human ear after making a call Perceive the quality of voice. In 1996, the international ITU organization began to develop relevant evaluation criteria in ITU-T Recommendations P.800 and P.830: Mean Opinion Score (MOS), and the average subjective opinion is a test method for subjective evaluation. Investigate and quantify the behavior of user listening and perceived voice quality, and the different survey users will separately focus on the original standard voice and the decaying sound after being transmitted through the wireless network.

Compare the feelings and compare the MOS scores. Table 1 is the MOS score comparison table:

level	MOS score	customer satisfaction
excellent	5.0	Very good, very clear, no distortion, no delay
good	4.0	Slightly poor, clearly heard, delayed, a little murmur
in	3.0	OK, I don’t know clearly, there is a certain delay, there is noise, there is distortion
difference	2.0	Reluctant, not very clear, with loud noise or intermittent, serious distortion
inferior	1.0	Very poor, silent or completely unclear, loud noise

Table 1

Referring to the MOS standard, the ITU organization has implemented a large number of end-to-end voice quality objective test technology standardization work, and released several voice evaluation algorithm standards: PAMS, Perceptual Voice Quality Measurement (PSQM), PSQM+, MNB, PESQ, and so on. MOS is a widely recognized voice quality standard. The above speech evaluation algorithm can ultimately correspond to the MOS.

The voice correlation parameters of two or more sets of test points pre-stored in the embodiment of the present invention, and the real voice quality corresponding to the voice correlation parameters of each set of test points can be implemented by: establishing a voice database of test points, voice The database stores the voice correlation parameters of multiple sets of test points and the corresponding real voice quality; optionally, the quality formula is used as the standard regression equation, and the preset evaluation parameters are used as the standard regression coefficients, wherein the true speech quality M' is the cause Variables, b1, b2, b3, and b4 are used as independent variables. The standard regression coefficient is solved by the least squares method on the premise that the sum of squared errors is required. Taking this embodiment as an example, the standard equations for solving the standard regression parameters are:

The values of a1, a2, a3, and a4 are obtained. It is worth noting that the multiple regression is a related art and includes many methods, and the embodiment is not limited thereto.

Optionally, after the step 102, the method of the embodiment of the present invention further includes: obtaining a real voice quality of the test point; determining whether a difference between the voice quality of the test point and the real voice quality is less than a preset threshold; and performing voice quality of the test point and When the difference of the real voice quality is greater than or equal to the preset threshold, the voice correlation parameter of the test point and the real voice quality of the test point are saved as a group to the voice database; according to the voice correlation parameter of each group of test points in the voice database The corresponding real speech quality and quality formula are subjected to multiple regression to obtain new evaluation parameters; the new evaluation parameters are used to update the preset evaluation parameters. Here, the embodiment may be a special process of updating the evaluation parameters, and the evaluation coefficients are repeatedly corrected to improve the accuracy of the evaluation coefficients. Taking the MOS score comparison as an example, the preset threshold may be equal to: the true voice quality multiplied by 2 percent, and the true speech quality multiplied by 2 percent is only a selectable value of the embodiment of the present invention, according to The requirements for test accuracy can be adjusted accordingly.

The embodiment of the invention further provides a computer storage medium, wherein the computer storage medium stores computer executable instructions, and the computer executable instructions are used to perform the method for determining the voice quality.

Embodiment 2

The embodiment of the invention provides a method for determining a voice quality, and the method is applied to a voice quality determining device, which may be a separate measuring device or a receiving device at the receiving end. As shown in FIG. 2, the method may include:

Step 201: Establish a voice database of test points.

Here, the voice database stores voice correlation parameters of multiple sets of test points and corresponding real voice quality. It should be noted that the voice database may include multiple sets of data measured by the researcher in advance through the measuring instrument to ensure that the preset evaluation parameters can be calculated.

Step 202: Determine a preset evaluation parameter of the test point according to the voice database.

Here, the preset evaluation parameters may include: a coverage correction coefficient a1, an interference correction coefficient a2, a transmission quality correction coefficient a3, a handover delay correction coefficient a4, and an environmental correction parameter c; therefore, at least five sets of the test points are stored in the database. The speech correlation parameter and the corresponding real speech quality can be used to calculate at least one set of evaluation parameters.

Optionally, the reference signal received power b1, the signal-to-interference plus noise ratio b2, the block error rate b3, the switching interval time b, and the real voice quality M′ of each group are substituted into a quality formula: M=a1* B1+a2*b2+a3*b3+a4*b4+c, to obtain multiple sets of equations, linearly transform these equations to obtain standard equations of standard regression parameters:

Solve a1, a2, a3, a4, and c as preset evaluation parameters.

Step 203: Acquire a reference signal received power b1 at the test point, a signal to interference plus noise ratio b2, a block error rate b3, and a switching interval time b4.

Step 204: Substituting b1, b2, b3, b4, and preset evaluation parameters into a quality formula, and determining a voice quality M of the test point. The preset evaluation parameters include a1, a2, a3, a4, and c described above.

Optionally, the mass formula is M=a1*b1+a2*b2+a3*b3+a4*b4+c.

Step 205: Determine, according to a preset rule, a level corresponding to M.

Here, the preset rule of the embodiment of the present invention may include a MOS score comparison table: the level corresponding to the voice quality is divided into five levels, as shown in Table 1. The value of M between 0 and 1 indicates a level that is completely inaudible. The value of M between 1 and 2 corresponds to level 2, which is unclear. The value of M is 2 to The corresponding level between 3 is the general level 3, the value of M is between 3 and 4, and the level corresponding to the level is 4, and the value of M is between 4 and 5. Level 5.

Step 206: When the level is 1 or 2, adjust b1, b2, b3, and b4.

Optionally, when the level representation is unclear or completely inaudible, the parameter with the largest weight is determined from b1, b2, b3, and b4, and the voice correlation parameter multiplied by the maximum weight parameter is increased. For example, when a1 is the largest among a1, a2, a3, and a4, b1 multiplied by a1 is increased, so that M can be increased and the quality level can be improved.

This embodiment may include two phases: one is a measurement phase, and the other is a correction phase;

The measurement phase includes the contents of steps 201 to 206, and when the M calculated by a1, a2, a3, a4, and c in the present embodiment is not so accurate, the correction phase is entered.

In the correction phase, the steps 203 to 205 are performed; at the same time, the actual voice quality M' is measured by the measuring instrument; whether the difference between the M and the M' is less than a preset threshold; if the difference between M and M' is greater than or equal to the preset Threshold, the speech correlation parameter and M' are saved as a set of data to the speech database, and multiple regressions are performed according to each set of speech correlation parameters, corresponding actual speech quality and quality formula in the speech database, and new evaluation parameters are obtained. The preset evaluation parameters are updated with new evaluation parameters; if so, the preset evaluation parameters are not changed. Here, in general, there is definitely an error between the calculated M and the measured M', and therefore, the preset threshold is the error of both.

In the embodiment of the present invention, taking the MOS score comparison as an example, the preset threshold may be equal to: the real voice quality multiplied by 2 percent, and the real voice quality multiplied by 2 percent is only one of the embodiments of the present invention. The values are selected and adjusted according to the requirements of the test accuracy.

Embodiment 3

The embodiment of the present invention provides a device 30 for determining a voice quality. As shown in FIG. 3, the device 30 may include:

The obtaining unit 301 is configured to acquire a voice correlation parameter of the test point in the current measurement when performing voice quality measurement.

The determining unit 302 is configured to determine the voice quality of the test point according to the acquired voice relevance parameter and the preset evaluation parameter affecting the voice.

In this way, during each test, it is no longer necessary to use the measuring instrument to measure the test point, but to estimate the voice quality of the test point according to the voice correlation parameter of the test point and the preset evaluation parameter. Since the measuring instrument is no longer needed, the measurement cost is greatly reduced.

Optionally, when the preset evaluation parameters affecting the voice include the voice evaluation parameter and the environment correction parameter, The determining unit 302 is set to:

The speech correlation parameter, the speech evaluation parameter and the environmental correction parameter are substituted into the quality formula to obtain the speech quality of the test point, and the quality formula is: M=A*B+c;

Where M is the speech quality of the test point; A is a 1×n matrix, including n speech evaluation parameters; B is an n×1 matrix, including n speech correlation parameters; c is an environmental correction parameter, and n is a positive integer .

Optionally, the voice evaluation parameter includes an overlay correction coefficient a1, an interference correction coefficient a2, a transmission quality correction coefficient a3, and a handover delay correction coefficient a4;

The voice correlation parameter includes a reference signal received power b1, a signal to interference plus noise ratio b2, a block error rate b3, and a handover interval time b4, and the handover interval time b4 is a time interval from the cut-in cell to the current measurement; the corresponding quality The formula is: M = a1 * b1 + a2 * b2 + a3 * b3 + a4 * b4 + c.

Optionally, as shown in FIG. 4, the apparatus 30 of the embodiment of the present invention further includes:

The first calculating unit 304 is configured to perform multiple regression according to the voice correlation parameters of the two or more sets of test points stored in advance, and the real voice quality and quality formula corresponding to the voice correlation parameters of each set of test points, to obtain a pre-regression Set evaluation parameters.

The voice correlation parameters of the two or more sets of test points stored in advance, and the real voice quality corresponding to the voice correlation parameters of each set of test points may be implemented by the establishing unit 303, and the establishing unit 303 is configured to establish a voice database, voice The database stores voice correlation parameters of multiple sets of test points and corresponding real voice quality;

Optionally, the obtaining unit 301 is further configured to obtain a real voice quality of the test point;

As shown in FIG. 4, the apparatus 30 of the embodiment of the present invention further includes:

The determining unit 305 is configured to determine whether the difference between the voice quality of the test point and the real voice quality is less than a preset threshold;

The saving unit 306 is configured to save the voice correlation parameter of the test point and the real voice quality of the test point as a set of data when the difference between the voice quality of the test point and the real voice quality is greater than or equal to the preset threshold.

It should be noted that the data save can be saved in the voice database established by the establishing unit 303.

The second calculating unit 307 is configured to perform multiple regression according to the voice correlation parameter of each group of test points in the voice database and the real voice quality and quality formula corresponding to the voice correlation parameter of the test point, to obtain a new evaluation parameter; The evaluation parameters update the preset evaluation parameters;

The update unit 308 is arranged to update the preset evaluation parameters with the obtained new evaluation parameters.

In an actual application, the obtaining unit 301, the determining unit 302, the establishing unit 303, the first calculating unit 304, the determining unit 305, the saving unit 306, the second calculating unit 307, and the updating unit 308 may each be a central processor located in the device 30. (Central Processing Unit, CPU), Microprocessor Unit (MPU), Digital Signal Processor (DSP), or Field Programmable Gate Array (FPGA).

Those skilled in the art will appreciate that embodiments of the present invention can be provided as a method, system, or computer program product. Accordingly, the present invention can take the form of a hardware embodiment, a software embodiment, or a combination of software and hardware. Moreover, the present invention can take the form of a computer program product embodied on one or more sets of computer usable storage media (including but not limited to disk storage and optical storage, etc.) containing computer usable program code.

The present invention has been described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (system), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or FIG. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing device to produce a machine for the execution of instructions for execution by a processor of a computer or other programmable data processing device. Means for implementing the functions specified in one or more sets of blocks and/or sets of blocks of a flow or a plurality of sets of blocks.

The computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture comprising the instruction device. The apparatus implements the functions specified in one or more sets of blocks of a flow or sets of flows and/or block diagrams of the flowchart.

These computer program instructions can also be loaded onto a computer or other programmable data processing device such that a series of operational steps are performed on a computer or other programmable device to produce computer-implemented processing for execution on a computer or other programmable device. Instructions are provided for implementation in the flowchart The steps of a function specified in a block or groups of blocks in a process or groups of processes and/or block diagrams.

The above description is only for the embodiments of the present invention and is not intended to limit the scope of the present invention.

Industrial applicability

The above technical solutions improve the efficiency of voice quality testing and reduce the measurement cost of voice quality testing.

Claims (10)

1. A method for voice quality determination, the method comprising:

   When performing voice quality measurement, obtain the voice correlation parameter of the test point in the current measurement;

   Determining the voice quality of the test point according to the obtained voice correlation parameter and the preset evaluation parameter affecting the voice.
2. The method according to claim 1, wherein the predetermined evaluation parameters affecting the voice include a voice evaluation parameter and an environment correction parameter;

   Determining, according to the obtained voice correlation parameter and the preset evaluation parameter that affects the voice, the voice quality of the test point includes:

   Substituting the obtained voice correlation parameter, the voice evaluation parameter and the environment correction parameter into a quality formula to obtain a voice quality of the test point, the quality formula is: M=A*B+c;

   Wherein said M is a speech quality of said test point; said A is a 1×n matrix comprising n speech evaluation parameters; said B being an n×1 matrix comprising n speech correlation parameters; c is an environmental correction parameter, and n is a positive integer.
3. The method according to claim 2, wherein said voice evaluation parameters include coverage correction coefficient a1, interference correction coefficient a2, transmission quality correction coefficient a3, and handover delay correction coefficient a4;

   The voice correlation parameter includes a reference signal received power b1, a signal to interference plus noise ratio b2, a block error rate b3, and a switching interval time b4, which is a time interval from the cut-in cell to the current measurement. ;

   The corresponding mass formula is: M = a1 * b1 + a2 * b2 + a3 * b3 + a4 * b4 + c.
4. The method according to any one of claims 1 to 3, before the obtaining the voice correlation parameter of the test point in the current measurement, the method further includes:

   Performing multiple regression according to the voice correlation parameters of the two or more sets of the test points stored in advance, and the real voice quality corresponding to the voice correlation parameter of each set of the test points and the quality formula, and obtaining the prediction Set evaluation parameters.
5. The method of claim 2 or 3, further comprising:

   Substituting the voice correlation parameter, the voice evaluation parameter and the environment correction parameter into the quality public After obtaining the speech quality of the test point,

   Obtaining the true voice quality of the test point;

   Determining whether a difference between the voice quality of the test point and the real voice quality is less than a preset threshold;

   When the difference between the voice quality of the test point and the real voice quality is greater than or equal to a preset threshold, the voice correlation parameter of the test point and the real voice quality of the test point are saved as a set of data;

   Performing multiple regression according to the voice correlation parameter of each set of the test points, the real voice quality corresponding to the voice correlation parameter of the test point, and the quality formula, to obtain a new evaluation parameter;

   The preset evaluation parameter is updated with the obtained new evaluation parameter.
6. A device for determining voice quality, the device comprising:

   The obtaining unit is configured to obtain a voice correlation parameter of the test point in the current measurement when performing voice quality measurement;

   The determining unit is configured to determine the voice quality of the test point according to the obtained voice relevance parameter and the preset evaluation parameter that affects the voice.
7. The apparatus according to claim 6, wherein the predetermined evaluation parameter affecting the voice comprises a voice evaluation parameter and an environment correction parameter, and the determining unit is configured to:

   Substituting the voice correlation parameter, the voice evaluation parameter and the environment correction parameter into a quality formula to obtain a voice quality of the test point, the quality formula is: M=A*B+c;

   Wherein said M is a speech quality of said test point; said A is a 1×n matrix comprising n speech evaluation parameters; said B is an n×1 matrix comprising n speech correlation parameters, c is an environmental correction parameter, and n is a positive integer.
8. The apparatus according to claim 7, wherein

   The speech evaluation parameter includes an overlay correction coefficient a1, an interference correction coefficient a2, a transmission quality correction coefficient a3, and a handover delay correction coefficient a4;

   The voice correlation parameter includes a reference signal received power b1, a signal and interference plus noise ratio b2, a block error rate b3, and a handover interval time b4, which is from the cut-in cell to the present Time interval between measurements;

   The corresponding mass formula is: M = a1 * b1 + a2 * b2 + a3 * b3 + a4 * b4 + c.
9. The device according to any one of claims 6 to 8, the device further comprising:

   The first calculating unit is configured to perform, according to the voice correlation parameter of the two or more sets of the test points stored in advance, and the real voice quality and the quality formula corresponding to the voice correlation parameter of each group of the test points. Multiple regression, the preset evaluation parameters are obtained.
10. A device according to claim 7 or 8,

    The obtaining unit is further configured to acquire a real voice quality of the test point;

    The device also includes:

    The determining unit is configured to determine whether a difference between the voice quality of the test point and the real voice quality is less than a preset threshold;

    The saving unit is configured to: when the difference between the voice quality of the test point and the real voice quality is greater than or equal to a preset threshold, the voice correlation parameter of the test point and the real voice quality of the test point are taken as one Group data is saved;

    The second calculating unit is configured to perform multiple regression according to the voice correlation parameter of each set of the test points, the real voice quality corresponding to the voice correlation parameter of the test point, and the quality formula, to obtain a new evaluation parameter;

    The update unit is configured to update the preset evaluation parameter with the obtained new evaluation parameter.

Images