CN101662550B - Method and system for service quality detection for call center - Google Patents

Method and system for service quality detection for call center Download PDF

Info

Publication number
CN101662550B
CN101662550B CN 200910170363 CN200910170363A CN101662550B CN 101662550 B CN101662550 B CN 101662550B CN 200910170363 CN200910170363 CN 200910170363 CN 200910170363 A CN200910170363 A CN 200910170363A CN 101662550 B CN101662550 B CN 101662550B
Authority
CN
China
Prior art keywords
service
recording
user
quality
parameter
Prior art date
Application number
CN 200910170363
Other languages
Chinese (zh)
Other versions
CN101662550A (en
Inventor
赖永森
郭潺
Original Assignee
中兴通讯股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中兴通讯股份有限公司 filed Critical 中兴通讯股份有限公司
Priority to CN 200910170363 priority Critical patent/CN101662550B/en
Publication of CN101662550A publication Critical patent/CN101662550A/en
Application granted granted Critical
Publication of CN101662550B publication Critical patent/CN101662550B/en

Links

Abstract

The invention discloses a method and a system for service quality detection for a call center. The method comprises the following steps of: obtaining service sound recording of the call center; extracting the attribute information in the service sound recording; determining a parameter value used for quality detection according to the attribute information and the corresponding relationship between the attribute information and a parameter for quality detection; and determining the service quality assessment value of the telephone traffic service according to the parameter value. The method can objectively detect the service quality of the telephone traffic service each time and has high realizing efficiency and high universality.

Description

呼叫中心服务质量检测方法及系统 Call Center Service Quality detection method and system

技术领域 FIELD

[0001] 本发明涉及通信技术领域,尤指一种用于呼叫中心系统的话务员服务质量自动检测方法及系统。 [0001] The present invention relates to communication technologies, especially the quality of the service operator and a system for automatic detection method for a call center system.

背景技术 Background technique

[0002] 呼叫中心作为提升客户服务水平和沟通效率的基本手段,最初是由大型电信运营商和银行等高端行业的客户服务、大型的外包呼叫中心发展而来。 [0002] Call Center as a basic tool to enhance customer service levels and communication efficiency, originally developed by large telecom operators and banks and other high-end customer service industry, a large call center outsourcing evolved. 随着现代商业运营模式的不断转变,呼叫中心已经开始从原来呼入(Inbound)呼叫中心向呼出(Outbound)呼叫中心发展,也就使得呼叫中心从原来的成本中心向利润中心转变。 With the ever-changing modern business model, the call center has started from the original incoming calls (Inbound) to outbound call center development (Outbound) call center, call center also makes the transition from the original cost center to a profit center. 但是由于呼叫中心生产的“产品”(一般为话务员的服务语音)的特殊性,也就让呼叫中心在产品质量检测问题上面临着更高的难度。 However, due to the call center production of "product" (usually as an operator of voice services) are unique, allows call center in product quality testing problems facing higher degree of difficulty.

[0003] 现有的呼叫中心质检系统,一般是采用人工操作的方式完成,由质检人员抽取话务员的服务录音来进行分析、评定,例如:为话务员的服务打分等。 [0003] Existing call center systems quality, it is generally used to complete the manual operation mode, service quality control personnel extracted by the operator to analyze the recording, evaluation, for example: the score for the service operator. 这种人工抽取检测的方式,需要投入大量的人力、物力、财力来实现对话务员服务电话的检测。 This artificial extraction detection mode, you need to invest a lot of manpower, material and financial resources to achieve the detection of operator services calls. 事实上,质检人员不可能对每一个服务电话进行监听分析,也就不可能有足够的时间去完成对话务员全部服务情况的了解,导致质检结果的片面性和不完整性,再加上质检人员主观评判等因素,从而使得质量检测结果会不准确、不公正。 In fact, quality control personnel is not possible to monitor every service call analysis, it is impossible to have enough time to complete all the knowledge of the operator of service, resulting in one-sided and incomplete quality control results, coupled with the quality seized subjective judgment and other factors, so that the quality test results will be inaccurate and unfair.

发明内容 SUMMARY

[0004] 本发明实施例提供一种呼叫中心服务质量检测方法及系统,实现对呼叫中心服务质量的自动检测,检测结果客观。 Embodiment [0004] The present invention provides a call center service quality detection method and system for automatic detection of the call center service quality, and objective detection result.

[0005] 一种呼叫中心服务质量检测方法,包括: [0005] A call center services quality inspection method, comprising:

[0006] 获取呼叫中心的服务录音; [0006] Gets the call center service record;

[0007] 提取所述服务录音中的属性信息,所述属性信息包括下列信息中的至少两种:月艮务时长、用户录音的性别及情感和话务员录音的话务内容; [0007] extract the attribute information of the service record, the attribute information comprises at least two of the following information: that works to months duration, gender and emotional content traffic and attendant recording recording user;

[0008] 根据所述属性信息,及所述属性信息与质量检测的参数的对应关系,确定用于质量检测的参数值; [0008] The corresponding relationship between the attribute information and the attribute information quality parameter detected, determining a value of the quality parameter detected;

[0009] 所述确定用于质量检测的参数值包含下列确定过程中的至少两种: [0009] The parameter value for determining the mass comprising the detection of at least two of the determination process:

[0010] 确定所述服务录音的服务时长,根据设置的服务时长阈值区间和服务时长参数的对应关系,查找获取到的服务录音的服务时长所在的阈值区间,确定服务录音的服务时长参数值; [0010] determining duration of the service record of the service, the threshold interval length located According to a service provided in corresponding relation size parameter duration threshold value interval and services, find the acquired service record of the service, determine the length of the parameter value at the service record of the service;

[0011] 确定出所述服务录音中包含的话务员录音,根据所述话务员录音的话务内容以及预先设定话务内容与内容质量参数的对应关系,确定对应的内容质量参数值; [0011] Recording of the service operator is determined that the recording contains, according to the traffic of the operator content and the correspondence between the recording preset traffic content and content quality parameter is determined corresponding to the content quality parameter value;

[0012] 确定出所述服务录音中包含的用户录音,根据所述用户录音所对应的录音用户的性别和情感,以及预先设置的不同性别的各种情感与用户情感参数的对应关系,确定对应的用户情感参数值;[0013] 根据所述确定出的参数值,确定本次话务服务的服务质量测评值。 [0012] determined that the user of the service record contained in the recording, according to the user record corresponding to recording user's gender and emotion, and the correspondence between different emotions and sex of user emotion parameter set in advance, to determine the corresponding the user emotion parameter; [0013] the determined values ​​of the parameter, determining the service quality evaluation value of the current traffic and services.

[0014] 一种呼叫中心服务质量检测系统,包括: [0014] A call center services quality inspection system, comprising:

[0015] 录音获取模块,用于获取呼叫中心的服务录音; [0015] Recording acquiring module, for acquiring the call center service record;

[0016] 提取模块,用于提取所述服务录音中的属性信息,所述属性信息包括下列信息中的至少两种:服务时长、用户录音的性别及情感和话务员录音的话务内容; [0016] extraction means for extracting attribute information of the service record of the attribute information comprises at least two of the following information: service duration, gender and emotional content traffic and attendant recording recording user;

[0017] 参数确定模块,用于根据所述属性信息,及所述属性信息与质量检测的参数的对应关系,确定用于质量检测的参数值;所述参数确定模块具体包括:时长分析子模块、确定子模块、质量分析子模块,以及情感分析子模块;所述时长分析子模块用于确定所述服务录音的服务时长,根据设置的服务时长阈值区间和服务时长参数的对应关系,查找获取到的服务录音的服务时长所在的阈值区间,确定服务录音的服务时长参数值;所述确定子模块,用于分别确定出所述录音获取模块获取到的服务录音中包含的话务员录音和用户录音;所述质量分析子模块,用于根据所述话务员录音的话务内容以及预先设定话务内容与内容质量参数的对应关系,确定对应的内容质量参数值;所述情感分析子模块,用于根据所述用户录音所对应的录音用户的性别和情感,以及 [0017] The parameter determination module, according to a corresponding relationship between the attribute information and the attribute information of the quality of the detected parameters, the parameter value for determining the quality testing; the parameter determining module comprises: a long analysis submodule , determining submodule, quality analysis sub-module, and emotional analysis sub-module; the duration analysis sub-module for determining the duration of the service record of the service, based on the correspondence size parameter duration threshold value interval and the service when the service is provided, the lookup to obtain the threshold value interval when the service record of the service length where determined length parameter values ​​services recording services; the determining sub-module, for respectively determining that the recording acquiring operator recording and user recording module acquires service record contained ; the quality analysis sub-module, according to the traffic of the operator content and the correspondence between the recording preset traffic content and content quality parameter, determining the quality parameter value corresponding to the content; the sentiment analysis sub-module, with the user at the user record corresponding to the recording and emotional sex, and 预先设置的不同性别的各种情感与用户情感参数的对应关系,确定对应的用户情感参数值; Various preset corresponding relation of emotion and gender user emotion parameter, determining a parameter value corresponding to the user emotion;

[0018] 质量确定模块,用于根据所述参数值,确定本次话务服务的服务质量。 [0018] The quality determination module, according to the parameter value for determining the quality of service of the current service traffic.

[0019] 本发明实施例提供的呼叫中心服务质量检测方法及系统,通过对获取呼叫中心的服务录音的至少两种属性信息,例如:服务时长、用户录音的性别及情感和话务员录音的话务内容等进行综合评定,根据预先设定的测评规则,确定综合评定的测评值,由于对所有的服务录音可以采用相同的标准评定,其测评结果客观;且可以自动实现,其实现效率高;可以针对不同的业务设定不同的测评规则,其灵活性高和通用性强。 [0019] Call center quality of service detection method and system according to the embodiment of the present invention, for at least two attribute information obtaining service recording call center, for example: when the service length, sex and emotional and attendant recording user recording traffic content comprehensive evaluation, the evaluation according to a predetermined rule, the value of comprehensive evaluation of the evaluation to determine, since all the services can be recorded using the same evaluation criteria, objective evaluation results thereof; and can be automated, which achieve high efficiency; can set different rules for different business evaluation, its high flexibility and versatility.

[0020] 上述方法充分考虑了服务时长、用户的情感和话务员的服务内容质量;通过从话务员和用户双方对话中,自动挖掘未经结构化的音频数据,获取表达内容和表达模态(即用户情感),对话务员的内容质量进行要求,同时对用户在服务中的感觉进行了分析,在理解用户感受的基础上,加深对用户的了解,以提高在服务中服务水平;也就是说,该测评结果是综合考虑了话务员的服务和被服务用户的感受,能够实现对话务服务的全面评价。 [0020] The above-described method fully considered long service, the service quality of the content emotional and the operator of the user; by the operator and the user sessions both in automatic drill unstructured audio data, obtaining the expression contents and expression of modality (i.e., a user emotion), the contents of the quality of the operator will be required, while the user feels in the service were analyzed, in understanding the user experience, based on a better understanding of the user, in order to improve service levels in the service; that is, the considering the survey results is the operator of the service and the served user experience, to achieve a comprehensive assessment to the traffic services.

[0021] 上述方法的实现可以完全由计算机等具有处理功能的设备、装置或系统自动完成,不需要人工参与,加快了呼叫中心质量检测的执行速度,降低了成本投入、提高了监控效率。 [0021] The above method can be implemented entirely by a device such as a computer having a processing function, apparatus or system automatically, without human involvement, faster execution quality testing of the call center, drive down the cost and improve the efficiency of monitoring.

附图说明 BRIEF DESCRIPTION

[0022] 图I为本发明实施例中呼叫中心服务质量检测方法的流程图; [0022] Figure I is a flowchart of call center services quality inspection method of the invention embodiment;

[0023] 图2为本发明实施例一中呼叫中心服务质量检测方法的具体实现流程图; [0023] FIG 2 flow chart for implementing specific call center services of a quality inspection method in Examples of the present invention;

[0024] 图3为本发明实施例一中对话务员录音进行内容质量分析的流程图; [0024] In Example 3 a flowchart for the operator-content recording mass analysis embodiment of the present invention;

[0025] 图4为本发明实施例一中对用户录音进行情感分析的流程图; [0025] FIG 4 is a flowchart of an embodiment of the emotional analysis of the user recording embodiment of the invention;

[0026] 图5为本发明实施例中呼叫中心服务质量检测系统的结构示意图。 [0026] FIG. 5 is a schematic structure of the call center services quality inspection system according to the present invention.

具体实施方式 Detailed ways

[0027] 本发明实施例提供的呼叫中心服务质量检测方法,提取呼叫中心的每一次服务录音,根据每一次服务录音的服务时长参数、用户情感参数和内容质量参数,综合评定服务录音的服务质量。 [0027] Call center service quality inspection method according to an embodiment of the present invention, to extract every service record of the call center, size parameter, user emotion parameters and contents quality parameters, the service quality appraisal service record in accordance with each time service record service . 其流程如图I所示,即在实际检测过程中,针对每一次服务录音,执行步骤如下: The process shown in FIG I, i.e. in the actual detection process, for each service record, perform the following steps:

[0028] 步骤SI :犾取呼叫中心的服务录首。 [0028] Step SI: l is the first recorded take service call center.

[0029] 呼叫中心一般会设置坐席录音模块,用于对话务员的服务进行录音。 [0029] The call center agent typically is provided recording means for recording on the operator services. 所以,可以从录音数据库中获取到每一次话务服务的服务录音。 Therefore, the recording can be obtained from the database to record every service traffic services.

[0030] 步骤S2 :提取获取的服务录音中的属性信息。 [0030] Step S2: extract attribute information services recording the acquired.

[0031] 属性信息包括下列信息中的至少两种:服务时长、用户录音的性别及情感和话务员录音的话务内容。 [0031] The attribute information includes at least two of the following information: a service length, sex and emotional content traffic and attendant recording of your recording.

[0032] 步骤S3 :根据提取的属性信息,及属性信息与质量检测的参数的对应关系,确定用于质量检测的参数值。 [0032] Step S3: The corresponding relationship between the extracted attribute information and attribute information quality parameter detected, determining a quality parameter value detection.

[0033] 根据属性信息的种类,相应的,可以确定出至少两种用于质量检测的参数值。 [0033] The category attribute information, the corresponding, at least two parameters can be determined for values ​​of quality inspection.

[0034] 则根据提取的属性信息确定用于质量检测的参数值的过程,相应的包含下列确定过程中的至少两种: [0034] the process according to the attribute information extracted parameter value for determining the quality of the detected, corresponding determination process comprising at least two of the following:

[0035] (i)确定服务录音的服务时长,根据设置的服务时长阈值区间和服务时长参数的对应关系,查找获取到的服务录音的服务时长所在的阈值区间,确定服务录音的服务时长参数值。 [0035] (i) determining the length of the service record of the service, the threshold interval length is located according to a service provided in corresponding relation size parameter duration threshold value interval and services, find the acquired service record of the service, it is determined length parameter values ​​Services Recording Services .

[0036] (ii)确定出服务录音中包含的话务员录音,根据话务员录音的话务内容以及预先设定话务内容与内容质量参数的对应关系,确定对应的内容质量参数值; [0036] (ii) determining an operator service record contained in the recording, according to the traffic operator content and the correspondence between the recording of a preset traffic content and content quality parameter is determined corresponding to the content quality parameter value;

[0037] (iii)确定出服务录音中包含的用户录音,根据用户录音所对应的录音用户的性别和情感,以及预先设置的不同性别的各种情感与用户情感参数的对应关系,确定对应的用户情感参数值。 [0037] (iii) it is determined that the user recording comprises recording service according to the user record corresponding to the user's gender and recording emotions, and gender of the preset corresponding relationship between the emotions of the user emotion parameter, determining a corresponding user emotion parameter values.

[0038] 步骤S4 :根据确定出的用于质量检测的参数值,确定本次话务服务的服务质量测评值。 [0038] Step S4: according to the determined parameter values ​​for the quality inspection, to determine the current quality of service evaluation value of the traffic and services. 具体包括: Including:

[0039] 根据设置的确定出的参数值在服务质量测评值中各自所占的权重比例,计算本次话务服务的服务质量测评值。 [0039] The weight ratio of weights of the determined setting parameters in the evaluation of quality of service values ​​in the respective share, calculates evaluation values ​​of this quality of service traffic and services.

[0040] 实施例一 [0040] Example a

[0041] 下面以采用服务时长、用户录音的性别及情感和话务员录音的话务内容三种属性信息进行话务质量检测为例进行详细说明。 [0041] Below when using long service, traffic information content of three properties sex and emotional and attendant recording of your recording quality testing carried traffic for example described in detail. 其流程如图2所示,包括下列步骤: Which process shown in Figure 2, comprising the steps of:

[0042] 步骤SlO :获取呼叫中心的服务录音。 [0042] Step SlO: acquiring the service record of the call center.

[0043] 从录音数据库中获取到每一次话务服务的服务录音中包括话务员录音和用户录 [0043] retrieved from the database to record every service record traffic services included attendant recording and record user

曰◦ Said ◦

[0044] 当采用单线程进行录音时,话务员录音和用户录音是在一起的;当采用双线程进行录音时,对话务员和用户分别进行录音,则话务员录音和用户录音是两个独立的部分。 [0044] When recording a single-threaded, the operator and the user recording is recorded together; double thread when recording to an operator and each user record, users record and recording the attendant are two separate parts.

[0045] 较佳的,在获取到服务录音后,对服务录音进行去噪处理。 [0045] Preferably, after obtaining the service record, service record denoising. 例如:可以通过基于矩阵实验室(Matrix Laboratory, Mat I ab)的小波变换消噪方法,对获取到的服务录音进行消噪处理。 For example: can, for recording the acquired service processing performed by the wavelet transform denoising noise removal based on laboratory matrix (Matrix Laboratory, Mat I ab) is.

[0046] 基于Matlab进行Daubechies (dbN, N表不小波的阶数)小波降噪,通过设置该强制消噪方法中的阈值,将小波分解结构中的高频系数全部设置为0,即滤掉所有高频部分,然后对信号进行小波重构,以达到消除语音以外的高频噪音,获得较为平滑的服务录音的效果。 [0046] Based on the Matlab Daubechies (dbN, N Table order not wavelet) Wavelet Denoising, provided by the forced threshold denoising method, high frequency coefficients of all the wavelet decomposition structure is set to 0, i.e., filtered All the high frequency portion, then the signal wavelet reconstruction, in order to eliminate high frequency noise than speech, obtain the effect of a smoother service recording.

[0047] 步骤Sll :根据获取到的服务录音的服务时长,确定对应的服务时长参数值。 [0047] Step Sll: The service acquired when recording long service, determining parameter values ​​corresponding to the length of service.

[0048] 根据设置的服务时长阈值区间和服务时长参数的对应关系,查找获取到的服务录音的服务时长所在的阈值区间,确定服务录音的服务时长参数值。 [0048] The correspondence between a service set size parameter threshold interval duration and service interval length threshold where the lookup service to obtain service record, determining the length of service parameter value recording service.

[0049] 例如,可以预先设定每个服务时长阈值区间所对应的服务时长得分值,根据获取到的服务录音的服务时长,查找其所在的阈值区间,从而确定其对应的得分值,即服务时长参数。 Grow score duration threshold value interval corresponding to the time service [0049] For example, it can be preset for each service, according to long acquired service record of the service, to find it in the threshold value interval, thereby determining a score value corresponding, long-service parameters. 如:服务时长阈值区间60〜120秒,对应的得分为90分;120〜150秒对应的得分为95分;等等。 Such as: a service interval duration threshold 60~120 seconds, corresponding to a score of 90 points; 120~150 seconds corresponding to a score of 95 points; and the like.

[0050] 步骤S12 :分别确定出获取的服务录音中包含的话务员录音和用户录音。 [0050] Step S12: it is determined that each operator service recording and audio recording user acquired included.

[0051] 当上述获取的服务录音是采用单线程进行录音时,从服务录音中提取出话务员录音和用户录音。 [0051] When recording the acquired service recording is single-threaded, the operator is extracted from the service recording and audio recording user. 上述获取的服务录音是采用双线程进行录音时,则只需确定出其中包含的两部分录音,哪部分是话务员录音,那部分是用户录音,而无需再进行提取。 When recording the acquired service is the use of dual-thread recording, simply determine the two parts of recordings contained therein, which is part of the operator recording, that part of the user is recorded, and no longer need to be extracted.

[0052] 步骤S13 :对话务员录音进行服务质量分析,确定对应的内容质量参数值。 [0052] Step S13: the recording quality of service for the operator-analysis to determine the quality parameter value corresponding to the content.

[0053] 将话务员录音转化为录音文本,根据录音文本中包含的预先指定服务关键词占录音文本中的总词汇量的比例以及设置的比例区间与内容质量参数的对应关系,确定对应的内容质量参数值。 [0053] The attendant recording into the recording text, the proportion of the total vocabulary recording text and the ratio of interval setting corresponding relation to the content according to the recording quality parameters comprising a pre-designated text service keywords, content quality is determined corresponding to parameter values. 如图3所示,具体包括下列步骤: As shown in FIG. 3, comprises the following steps:

[0054] 步骤S131 :通过语音文本转换软件将话务员录音转化为录音文本(即TXT格式的文本)。 [0054] Step S131: through a speech to text software attendant recording into the recording text (i.e., text TXT format).

[0055] 步骤S132 :统计转换出的录音文本中包含的预先指定的服务关键词的数量。 [0055] Step S132: the number of statistical text converted by recording the service included in the pre-specified keywords.

[0056] 其中,预先指定的服务关键词可以包括:产品业务知识词汇、服务用语、沟通技巧词汇等等。 [0056] where pre-specified services Keywords can include: product knowledge of business vocabulary, language services, communication skills, vocabulary and so on. 例如:产品名称和产品信息的关键语句等产品业务知识词汇;礼貌用语、致谢语、致歉语、开头语、结束语等服务用语;服务禁语,沟通技巧方面的引导用语、沟通用语等沟通技巧词汇。 For example: the product name and the key sentence of product information and other products business knowledge vocabulary; polite language, thanks language, apology language, chapeau, conclusion and other language services; services vow of silence, guide language skills of communication, communication language and other communication skills vocabulary .

[0057] 较佳的,可以针对不同的业务指定不同的服务关键词,该指定的服务关键词可以存储在知识库中,以备统计时使用。 [0057] Preferably, may be different for different traffic service specified keywords, the keywords may be designated service stored in the knowledge base, when used to prepare statistics.

[0058] 步骤S133 :统计转换出的录音文本中包含的词汇总量。 [0058] Step S133: convert the total vocabulary of statistical recordings contained in the text.

[0059] 步骤S134 :计算统计出的服务关键词的数量占录音文本中的词汇总量的比例。 [0059] Step S134: calculate the statistics of the number of service keywords proportion of the total vocabulary of recording text. 即计算得到服务关键词出现的概率,以便根据该出现概率确定内容质量。 That calculated the probability of occurrence of keyword services, in order to determine the quality of the content based on the probability that appears.

[0060] 步骤S135 :根据设置的比例区间与内容质量参数的对应关系,确定本次服务录音的内容质量参数值。 [0060] Step S135: The corresponding relationship between the content ratio of the mass of interval setting parameters, determining the content of this service quality parameter value recording.

[0061] 例如,可以预先设置若干比例区间及其对应的内容质量得分值,根据计算出的服务关键词的数量占录音文本中的词汇总量的比例值所在的比例区间,查找对应的内容质量得分值。 [0061] For example, the ratio may be preset intervals and a plurality of values ​​corresponding to the content quality score, the proportion of the total vocabulary recording interval ratio text where the value calculated according to the number of service keyword, find the corresponding content quality scores. 如:比例区间50-60%,对应的得分为90分;比例区间60-70%对应的得分为95 Such as: the ratio range 50-60%, corresponding to a score of 90; range 60-70% proportion corresponding to a score of 95

分;等等。 Points; and so on.

[0062] 步骤S14 :对用户录音进行情感分析,确定对应的用户情感参数值。 [0062] Step S14: Recording the user emotional analysis to determine the values ​​of the emotion parameters corresponding to the user.

[0063] 根据预先确定的性别概率阈值以及计算出的用户录音基音信号累积密度函数值,确定用户录音所对应的录音用户的性别;然后,根据录音用户的性别计算出的用户录音的情感特征参数以及统计出的不同性别各种情感所对应的情感特征参数分布区间,确定该录音用户的情感,例如:愤怒、恐惧、高兴、平静、悲伤、惊奇等情感;再根据确定出的用户录音的性别和情感,查找预先设定的不同性别的各种情感与用户情感参数的对应关系,确定该用户录音的用户情感参数值。 [0063] cumulative density function values ​​based on gender probability threshold determined in advance, and a user recording pitch calculated signal determining recording user's gender user record corresponding; then, affective characteristic parameters calculated from the recorded user's gender user recording and emotional parameters of the statistics of the distribution section of various emotions corresponding to different genders, and recording the user's emotion is determined, for example: anger, fear, happy, calm, sad, surprised like emotions; re-recording according to the determined user's gender and emotional, to find the corresponding relationship between gender and emotions user emotion parameter set in advance, determines user emotion parameter value of the user recording.

[0064] 其中,用户情感参数也可以通过情感得分值来体现,例如:预先设定平静对应的情感得分值为80,高兴对应的得分值为90等。 [0064] wherein the user emotion parameter can also be embodied by emotion score value, for example: calm predetermined value corresponding to the emotion score 80, a score value corresponding to the happy 90 and the like.

[0065] 步骤S15 :根据确定出的服务时长参数值、内容质量参数值和用户情感参数值,确定本次话务服务的服务质量测评值。 [0065] Step S15: The length parameter values ​​determined service quality parameter value and the content user emotion parameter value, determining the service quality evaluation value of the current service traffic. 具体可以采用设定的测评规则,确定本次话务服务的服务质量。 Specific evaluation can be used to set rules to determine the quality of service of this traffic services.

[0066] 根据上述得到服务时长参数值、内容质量参数值和用户情感参数值,以及测评规则中设定的各自的权重比例,计算得到本次话务服务的最终服务质量的测评值。 [0066] The length of service parameter values ​​obtained above, the contents of user emotion parameter value and quality parameter values, and the weight ratio of the respective weights are in the rule evaluation, the calculated value of the final evaluation of the quality of service of the current service traffic. 即计算服务时长参数值、内容质量参数值和用户情感参数值与各自对应的权重比例的乘积,并计算得到的各乘积之和,得到本次话务服务的服务质量的测评值。 Length computing service parameter values ​​i.e., the weight percentage of the product weight and the content of user emotion quality parameter value corresponding to each parameter value and the calculated sum of the multiplication to obtain the evaluation value of the current quality of service traffic and services. [0067] 例如:服务时长参数值、内容质量参数值和用户情感参数值所占权重比例为1:2: 2,则可以根据权重比例计算服务质量的测评值如下: [0067] For example: length parameter, quality of content a user emotion parameter values ​​and parameter values ​​of the service weight weight proportion of 1: 2: 2, may be as follows according to the weight ratio calculated evaluation value of the quality of service:

[0068] [0068]

测评值=服务时长参数XI +内容质量参数X鲁+用户情感参数X昏; Evaluation value = length parameter XI + content quality parameters Lu X + X user emotion parameter faint when services;

[0069] 本发明的上述呼叫中心服务质量检测方法,还包括: [0069] The call center services quality inspection method of the invention, further comprising:

[0070] 步骤S16 :获取话务员进行话务服务的操作信息。 [0070] Step S16: The operator operates acquiring traffic information and services.

[0071] 从呼叫中心的业务系统中获取存储的话务员进行话务服务的操作信息,操作信息中包含有每个坐席上的话务员的操作记录,根据操作记录可以确定出每次服务录音对应的话务员。 [0071] retrieving stored call center operator from the service system for the service traffic operation information, the operation information includes record of each operation of the operator on the seat, each service may be determined corresponding to the recording operation of a recording operator .

[0072] 步骤S17 :根据获取到的话务服务的操作信息,确定与获取的服务录音对应的执行本次话务服务的话务员。 [0072] Step S17: The operation information acquired traffic and services, the operator determines to perform the current traffic and services corresponding to the acquired service recording.

[0073] 即可以将每次获取的服务录音与进行操作的话务员进行关联,建立起关联关系。 [0073] which can be acquired each service record to associate with the operator to operate and establish relationships.

[0074] 步骤S18 :确定出每次话务服务的服务质量后,将确定出的服务质量测评结果进行存储。 [0074] Step S18: After each quality of service is determined that the traffic and services, the determined quality of service evaluation result is stored. 以便后续需要时调用。 The need for subsequent calls.

[0075] 该步骤可以在步骤S15之后执行,即存储步骤S15中得到每次话务服务的服务质量,实现可以查询、提取、每次服务录音的服务质量。 [0075] This step may be performed after the step S15, that is stored in step S15 get quality service every time traffic services, implementation can query, extract, each service quality service recording.

[0076] 也可以在步骤S17之后执行,即存储步骤S15中得到每次话务服务的服务质量测评值以及步骤S17中确定的与获取的服务录音对应的执行本次话务服务的话务员。 [0076] may be performed after step S17, step S15, i.e., storing the operator for executing the obtained traffic and services corresponding to each service record acquired evaluation value and the quality of service traffic service in step S17 determined. 从而可以实现查询、提取、统计存储的某个话务员的所有服务录音的服务质量,以便对每个话务员的服务质量进行全面、客观的评定。 Thereby enabling query, extract, service quality recordings of all the services of a storage operator statistics in order to conduct a comprehensive and objective assessment of the quality of service for each attendant.

[0077] 上述步骤S14对用户录音进行情感分析以确定对应的用户情感参数,其具体实现流程如图4示,包括下列步骤: [0077] The step S14 of recording the user emotional analyzed to determine the emotion parameters corresponding to the user, the specific implementation process is shown in FIG. 4, comprising the steps of:

[0078] 步骤S141 :根据预先确定的性别概率阈值以及计算出的所述用户录音基音信号累积密度函数值,对用户录音进行录音用户性别判断,确定用户录音所对应的录音用户的性别。 [0078] Step S141: The cumulative probability threshold sex predetermined recording pitch signal and the user density function value is calculated, the user record user recording sex determination, the user determines the sex of the user record corresponding to the record.

[0079] 可以利用不同性别的基音信号频率统计模型的差异对性别进行分离。 [0079] The difference of the pitch frequency of the signal statistical models may utilize different sex gender separation. [0080] 首先,需要计算用户录音的基音信号在累积函数曲线上性别的概率差最大的频率点的累积密度函数值P (fT)。 Cumulative density function values ​​[0080] First, the user needs to calculate the pitch signal recorded on the gender difference in the cumulative probability function curve of the maximum frequency point P (fT).

[0081] 然后,根据用户录音的基音信号累积密度函数值进行录音用户性别判断,具体为:将计算得到的累积密度函数值,与预先确定出的累积分布阈值进行比较,确定用户录音所对应的录音用户的性别。 [0081] Then, recording the user's gender determined according to the cumulative density function value of the user recording the fundamental tone signal, in particular: the cumulative density function value is calculated, and determined in advance cumulative distribution threshold value, it determines that the user record corresponding to recording the user's gender. 即:若得到的累积密度函数值大于累积分布阈值,则确定用户录音所对应的录音用户的性别为男性;否则,确定用户录音所对应的录音用户的性别为女性。 That is: If the cumulative density function value obtained is greater than the cumulative distribution threshold value, determining the sex of the user record user record corresponding male; otherwise, determining the sex of the user record corresponding to the user record for women.

[0082] 其中,累积分布阈值根据对不同性别的用户的基音信号在累积分布函数曲线上性别的概率差最大的频率点的累积密度函数值的统计结果确定。 [0082] wherein the threshold cumulative distribution statistics cumulative density function values ​​of the maximum frequency point is determined according to the user's gender pitch gender difference signal probability on cumulative distribution function curve. 例如:可以确定累积分布阈 For example: the cumulative distribution of the threshold may be determined

值为-Pr= m-.....-即选取累积分布函数曲线上性别的概率差最大的频率点4作 -Pr = m -.....- probability value that is selected on the cumulative distribution function curve of the maximum frequency difference between gender for the point 4

L • L •

为阀值频率,选取阈值频率对应的男女性别的累积密度函数值的均值作为累积分布阈值。 Is a threshold frequency, mean cumulative density function value to select the gender threshold frequency corresponding to the threshold value as a cumulative distribution.

[0083] 其中,fT为累积函数曲线上性别的概率差最大的频率点; [0083] wherein, fT is the cumulative probability function curve gender difference between the maximum frequency points;

[0084] Pffl(fT)为男性的基音信号累积分布函数; [0084] Pffl (fT) as the pitch of the signal male cumulative distribution function;

[0085] Pf(fT)为女性的基音信号累积分布函数。 [0085] Pf (fT) signal pitch female cumulative distribution function.

[0086] 例如:上述累积分布函数具体可以通过下列计算Parzen窗法进行计算得到,其过程如下: [0086] For example: the above-described specific cumulative distribution function may be calculated by the following calculation Parzen window method, which is as follows:

[0087] 针对设定区域9^的,基音信号概率密度函数可以通过下列公式(I)计算: [0087] ^ 9 for the set area, the pitch signal probability density function can be calculated by the following formula (I):

[0088] [0088]

n KIN n KIN

PN ( f) = N — PN (f) = N -

Vn Vn

[0089] 其中,^表示基音信号的概率密度; [0089] where ^ represents the probability density of the pitch signal;

[0090] Vn为设定区域Wn的覆盖范围;其中,《N是一个包含f在内的一个设定区域; [0090] Vn is set Wn of the coverage area; wherein, "N containing f is a region including a set;

[0091] N是所有样本个数; [0091] N is the number of all samples;

[0092] Kn为落入设定区域9^中的样本个数。 [0092] Kn is the number of samples falling within the set region 9 ^.

[0093] 落入区域9^的样本个数Kn可以通过定义的一个窗函数P(/)来表示,则根据Parzen窗估计法,即令f = (f1; f2,…,fn) S定义一个窗函数K/): Sample number Kn [0093] ^ 9 falls within the region may be represented by a defined window function P (/), the estimation method according to the Parzen window, and even if f = (f1; f2, ..., fn) S define a window function K /):

[0094] [0094]

•S 'tr2 • S 'tr2

[0095]其中,i = l,2,…,n。 [0095] where, i = l, 2, ..., n.

[0096] 实际上</)是以函数值I界定了一个以原点为中心,棱长为I的n维超立方体。 [0096] Indeed </) is a function of a value I defined centered at the origin, the edge length of I n-dimensional hypercube. 通过坐标的平移和尺度的缩放可以改变超立方体的位置和大小。 Hypercube can change the position and size by scaling the coordinates of the translation and scale. 如果一个样本A落入以f为中心以1%为棱长的区域内时则计数为1,否则计数为O。 If A falls within a sample to f is 1% of the center area of ​​the edge length of the count is 1, otherwise the count is O. 可以利用窗函数舛/)实现上 May be mistaken by using a window function /) implemented on

述约定,即: He said convention, namely:

[0097] [0097]

(Pif) = (P^l) (Pif) = (P ^ l)

K K

[0098] 则落入区域Ww的样本个数Kn的计算公式(2)为: [0098] calculated number of samples of Kn is Ww region falls (2):

[0099] [0099]

Figure CN101662550BD00101

[0100] 其中,i为正整数; [0100] where, i is a positive integer;

[0101] fi为定义的窗函数中的某个样本点的值; [0101] the value of a sample point is defined by a window function Fi;

[0102] hN表示正态窗函数窗体的棱长。 [0102] hN edge length indicates the normal form of the window function.

[0103] 使用概率密度函数正态窗函数炉(/),其计算公式(3)如下: [0103] normal probability density function using the window function furnace (/), the calculation formula (3) as follows:

[0104] [0104]

Figure CN101662550BD00102

[0105] 将公式(3)代入公式(2),得到落入区域Wv的样本个数Kn的计算公式⑷如下: [0105] Equation (3) into equation (2), is calculated to obtain the number of sample ⑷ Kn Wv region falls below:

Figure CN101662550BD00103

[0107] 将公式(4)代入公式(I),可以得到基音信号的概率密度函数的具体计算公式(5) Specific formula (5) [0107] Equation (4) into the formula (I), can be obtained of the probability density function of the fundamental tone signal

如下: as follows:

[0108] [0108]

Figure CN101662550BD00104

[0109] 公式(5)即为利用正态窗对基音频率的概率密度函数进行Parzen窗法估计的计算公式,利用公式(5)既可以对语音信号基音频率的概率密度进行估算;对该概率密度函数进行积分运算,即可得到基音频率累积分布函数。 [0109] Equation (5) is the window using normal probability density function of the pitch frequency estimate is calculated Parzen window method, using Equation (5) may be the pitch frequency of the speech signal probability density estimate; the probability density integration calculation function, the pitch frequency can be obtained by the cumulative distribution function.

[0110] 根据得到的基因频率累积分布函数,既可以计算频率点&对应的男性的基音信号累积密度函数值Pm(fT)和频率点&对应的女性的基音信号累积密度函数值Pf (fT);也可以计算用户录音的基音信号的累积密度函数值P (fT)。 [0110] The gene frequencies obtained cumulative distribution function, may be calculated frequencies & corresponding male pitch signal of the cumulative density function value Pm (fT) and frequencies & corresponding female pitch signal cumulative density function value Pf (fT) ; cumulative density function values ​​may be calculated pitch signal recording user P (fT).

[0111] 其中,男性的基音信号累积密度函数值Pm(fT)和频率点&对应的女性的基音信号累积密度函数值Pf (fT),可以通过对性别已知的录音样本数据的累积密度函数值进行统计分析确定。 [0111] wherein the male pitch signal cumulative density function value Pm (fT) and a frequency corresponding to point & female pitch signal cumulative density function value Pf (fT), known by gender data recorded sample CDF statistical analysis was performed to determine the value.

[0112] 步骤S142 :根据计算出的用户录音的情感特征参数、确定出的录音用户的性别以及统计出的不同性别各种情感所对应的情感特征参数分布区间,确定该录音用户的情感。 [0112] Step S142: The emotion feature parameters calculated from the user recording, the recording of the determined sex of the user, and the statistics of the different sex emotions corresponding to the emotional parameters of the distribution section, recording the user's emotion is determined. 具体包括: Including:

[0113] 其中,不同性别的各种情感所对应的情感特征参数分布区间,根据预先分类统计的各性别的不同情感的录音样本进行分类分析后确定的。 After [0113] wherein, gender emotions corresponding to the emotional parameters of the distribution section, classified according to the recording samples of different emotional analysis of the statistical classification of each sex in advance determined.

[0114] 具体可以采用基于情感的三维空间理论,对性别和情感已知的录音样本进行分类统计并存储,针对不同性别的各种情感的录音样本,例如:愤怒、恐惧、高兴、平静、悲伤、惊奇等,统计每种情感的情感特征参数所在的分布区间,从而确定不同性别的各种情感的情感特征参数的分布区间,例如:基音和共振峰的均值和方差分布区间。 [0114] Specific three-dimensional space based theory of emotion, gender and emotion known to classify statistics recorded sample and stored for the recorded sample a variety of different emotions of sex, such as: anger, fear, happiness, calm, sad , surprise, etc., the statistical distribution of each emotion emotional characteristics interval parameters is located, in order to determine the characteristic parameters of the distribution range of emotion emotions of different sex, for example: the mean and variance of the distribution range of pitch and formant.

[0115] 需要确定用户录音所对应的录音用户的情感时,计算用户录音的情感特征参数,并确定计算出的情感特征参数所在的分布区间,根据分布区间确定对应的情感即为该录音用户的情感。 When the [0115] recording user's emotion is determined by the user corresponding to the recording, calculating the emotional parameters of the user recording characteristic, and determining the distribution of the calculated interval where characteristic parameters of the emotions, emotion that is determined corresponding to the user based on the distribution record section emotion. 例如:计算用户录音的基音和共振峰的均值和方差,一般可以选择计算共振峰的前三阶的均值和方差;并查找计算出的用户录音的基音和共振峰的均值和方差所在的分布区间,根据分布区间确定对应的情感即为该录音用户的情感。 For example: a user computing pitch and formant recording means and variances, mean and variance can generally choose the first three formants calculated; distribution interval and find the pitch calculated from the user recording and the mean and variance of the resonance peak is located , determined according to the distribution interval is the emotion corresponding to the recorded user emotion.

[0116] 步骤S143 :根据确定出的用户录音的性别和情感,查找预先设定的不同性别的每种情感与用户情感参数的对应关系,确定该用户录音的用户情感参数值。 [0116] Step S143: determining an emotion based on gender and recording of the user, searches for the corresponding relationship between each emotion and the user emotion parameter preset sex determining user emotion parameter value of the user recording.

[0117] 例如:用户情感参数值具体可以通过用户情感得分值来表征。 [0117] For example: a user emotion parameter value may be characterized by a specific user emotion score value. 即可以预先设定每种情感所对应的用户情感得分值,根据确定出的用户情感,查找对应的用户情感得分值,例如:愤怒、恐惧、高兴、平静、悲伤、惊奇等用户情感分别对应的得分值。 I.e., a user emotion may be previously set point value corresponding to each emotion, according to the determined user emotion, to find a user emotion score values ​​corresponding to, for example: anger, fear, happy, calm, sad, surprised each other user emotion corresponding score values.

[0118] 根据本发明的上述呼叫中心服务质量检测方法,可以构建一种呼叫中心服务质量检测系统,如图5所示,包括:录音获取模块10、提取模块20、参数确定模块30和质量确定模块40。 [0118] The said call center service quality inspection method of the present invention can construct a call center service quality detection system, shown in Figure 5, comprising: recording an obtaining module 10, an extraction module 20, and a quality parameter determination module 30 determines module 40.

[0119] 录音获取模块10,用于获取呼叫中心的服务录音。 [0119] Recording obtaining module 10, configured to obtain the call center service recording.

[0120] 提取模块20,用于提取服务录音中的属性信息。 [0120] extraction module 20 for extracting the attribute information of service recording.

[0121] 提取模块20,具体用于提取服务录音的下列属性信息中的至少两种:服务时长、用户录音的性别及情感和话务员录音的话务内容。 [0121] extracting module 20, particularly for at least two of the extracted attribute information recorded in the service: service duration, gender and emotional content traffic and attendant recording user recording.

[0122] 参数确定模块30,用于根据提取模块20确定出的属性信息,及属性信息与质量检测的参数的对应关系,确定用于质量检测的参数值。 [0122] Parameter determination module 30, according to a correspondence relationship between the parameter extraction module 20 determines the attribute information and the attribute information detected by mass, mass parameter value for determining detection.

[0123] 相应的,参数确定模块30,具体包括:时长分析子模块301、确定子模块302、质量分析子模块303和情感分析子模块304。 [0123] Accordingly, the parameter determination module 30 comprises: a long analysis sub-module 301, the determination sub-module 302, quality analysis sub-module 303 and emotional analysis submodule 304.

[0124] 时长分析子模块301,用于确定服务录音的服务时长,根据设置的服务时长阈值区间和服务时长参数的对应关系,查找获取到的服务录音的服务时长所在的阈值区间,确定服务录音的服务时长参数值。 [0124] Duration analysis sub-module 301, for the duration of determining the service record of the service, based on the correspondence size parameter duration threshold value interval and the service when the service is provided, the threshold interval length where the lookup acquired service record of the service, determine the service record duration of service parameter values.

[0125] 确定子模块302,用于分别确定出录音获取模块获取到的服务录音中包含的话务员录音和用户录音。 [0125] determination sub-module 302, it is determined that recording for respectively recording and operator acquisition module acquires recording user service included in the recording.

[0126] 质量分析子模块303,用于根据话务员录音的话务内容以及预先设定话务内容与内容质量参数的对应关系,确定对应的内容质量参数值。 [0126] quality analysis sub-module 303, according to the traffic operator content and the correspondence between the recording of a preset traffic content and content quality parameter is determined corresponding to the content quality parameter value.

[0127] 较佳的,质量分析子模块303,进一步可以包括:转化单元3031、计算单元3032和质量确定单元3033。 [0127] Preferably, mass analysis sub-module 303 may further comprising: a conversion unit 3031, and a quality calculating unit 3032 determination unit 3033.

[0128] 转化单元3031,用于将话务员录音转化为录音文本。 [0128] conversion unit 3031, for converting the recording operator transcriptions.

[0129] 计算单元3032,用于计算统计出的所述录音文本中包含的预先指定的服务关键词的数量占统计出的所述录音文本中的词汇总量的比例值。 [0129] calculation unit 3032, a pre-specified number of services for calculating statistics of the keywords contained in the text of the recording proportion of the value of the total statistics of words in the transcriptions.

[0130] 质量确定单元3033,用于根据预先设定的比例区间与内容质量参数的对应关系,查找计算单元3032计算出的比例值所在的比例区间,确定本次服务录音对应的内容质量参数值。 [0130] quality determination unit 3033, according to a preset correspondence relationship between the content ratio range of quality parameters, to find the ratio calculation unit 3032 calculates the interval proportional value is located, to determine the content of the current service quality parameter value corresponding to the recording .

[0131] 情感分析子模块304,用于根据用户录音所对应的录音用户的性别和情感,以及预先设置的不同性别的各种情感与用户情感参数的对应关系,确定对应的用户情感参数值。 [0131] Emotion analysis sub-module 304, according to a user record corresponding to the user's gender recording and emotional, and the correspondence between different emotions and sex of user emotion parameter set in advance, determines a parameter value corresponding to the user emotion.

[0132] 较佳的,情感分析子模块304,进一步可以包括:性别确定单元3041、情感确定单元3042和参数确定单元3043。 [0132] Preferably, the emotional analysis sub-module 304 may further comprise: sex determination unit 3041, a determining unit 3042 and the emotion parameter determination unit 3043.

[0133] 性别确定单元3041,用于根据预先确定的性别概率阈值以及计算出的用户录音基音信号累积密度函数值,确定用户录音所对应的录音用户的性别。 [0133] sex determination unit 3041, a cumulative density function values ​​according to a predetermined probability threshold gender of the user and recording the calculated pitch signal, determining the sex of the user record corresponding to the user record. [0134] 情感确定单元3042,用于根据录音用户的性别、计算出的用户录音的情感特征参数以及预先统计出的不同性别各种情感所对应的情感特征参数分布区间,确定该录音用户的情感. [0134] emotion determination unit 3042, according to the distribution characteristic parameter emotional recording interval the user's gender, affective characteristic parameters calculated from the user in advance and recording the statistics of emotions corresponding to different genders to determine that the user's emotion recording .

[0135] 参数确定单元3043,用于根据确定出的用户录音的性别和情感,查找预先设定的不同性别的各种情感与用户情感参数的对应关系,确定该用户录音的用户情感参数值。 [0135] parameter determination unit 3043 for determining an emotion based on gender and recording user, searches for the corresponding relation set in advance gender of emotions and user emotion parameter, determining user emotion parameter value of the user recording.

[0136] 质量确定模块40,用于根据参数确定模块30确定出的参数值,确定本次话务服务的服务质量。 [0136] Quality determination module 40, according to parameter determination module 30 determines a parameter value, determining the current quality of service traffic and services. [0137] 质量确定模块40,具体用于:根据设置的参数值在服务质量测评值中各自所占的权重比例,计算本次话务服务的服务质量测评值。 [0137] Quality determination module 40, configured to: according to a weight ratio of weight parameter value set in each of the service quality evaluation value in proportion to calculate the current quality of service traffic and services evaluation value.

[0138] 上述系统还包括:消噪模块50,用于通过小波变换消噪对录音获取模块10获取的服务录音进行去噪处理。 [0138] The system further comprising: a noise cancellation module 50, configured to obtain service module 10 acquires the recording process of recording the denoising Denoising by wavelet transform.

[0139] 上述系统还包括:信息处理模块60和存储模块70。 [0139] The system further comprises: an information processing module 60 and a memory module 70.

[0140] 信息处理模块60,用于获取根据存储的话务员进行话务服务的操作信息,根据获取的操作信息确定与录音获取模块10获取的服务录音对应的执行本次话务服务的话务员。 [0140] The information processing module 60, configured to acquire operation information storage service according to the traffic of the operator, the operator obtaining module 10 obtains the corresponding service record for executing the traffic and services and recording the acquired operation information is determined in accordance with.

[0141] 存储模块70,用于将确定出的质量确定模块40确定出的服务质量测评值以及信息处理模块60确定出的执行本次话务服务的话务员进行存储。 [0141] a storage module 70 for determining the determined mass of the operator 60 is determined for executing the traffic and services module 40 determines an evaluation value of the quality of service and an information processing module for storage.

[0142] 较佳的,上述系统中还可以设置一个录音样本收集模块,为情感分析子模块304提供收集的录音不同性别的各种情感的录音样本。 [0142] Preferably, the above-described system may also be provided a recorded sample collection module, for the recording sentiment analysis sub-module 304 provides audio samples sex collected various emotions.

[0143] 本发明的上述呼叫中心服务质量检测系统,可以通过软件实现,也可以通过硬件实现,例如,可以是一种信息处理芯片。 [0143] The call center services quality inspection system according to the present invention, can be implemented by software, hardware, may be realized by, for example, may be an information processing chip.

[0144] 下面是一个具体的计算案例说明。 [0144] The following is a specific case of computational instructions.

[0145] 当一位男性用户打电话到呼叫中心进行查询某个业务信息的服务时,坐席录音模块会对该次话务服务进行录音。 [0145] When a user makes a call to a male call center service queries for business information, seating recording module to record the times traffic services. 录音获取模块可以从坐席录音模块中获取到该次的服务录音;较佳的,可以由消噪模块进行消噪处理。 Acquisition module may acquire from the recording agent to the recording module recording time service; Preferably, may be processed by the noise canceling noise canceling module.

[0146] 然后由时长分析子模块确定该次服务录音的服务时长参数值。 [0146] Then the length of the current service parameter values ​​determined by recording the duration of the service analysis submodule. 例如:本次话务服务的服务时长为70秒,设置的服务时长参数中,60〜120秒对应的得分值为90分,则本次的服务时长的得分即为90分。 For example: When the current service traffic and services to 70 seconds long, the size parameter, a score corresponding to 60~120 seconds is 90 minutes and the service provided, the length of time of this service is the score of 90 points.

[0147] 由确定子模块分别确定其中包含的话务员录音和用户录音。 [0147] recording and operator determines which comprises a recording of a user determination sub-module respectively. 然后,质量分析子模块对话务员录音进行服务质量分析,确定对应的内容质量参数值;例如:用户的情感为平静,而设置的不同性别各种情感所对应的情感特征参数分布区间中,性别为男性,情感为平静对应的得分值为80分,则本次话务服务的用户情感得分为80分。 Then, the operator of the quality analysis sub-module analyzes the recording quality of service, content quality parameter is determined corresponding to a value; for example: the user's emotion calm, and emotional parameters of emotions gender distribution section disposed in a corresponding, gender male, calm emotion corresponding score is 80 points, then this emotion traffic service user score of 80 points.

[0148] 情感分析子模块对用户录音进行情感分析,确定对应的用户情感参数值;例如:话务员录音中包含的服务关键词占其词汇总量的比例为18%,设置的服务关键词比例区间中,15%〜20%为80分,则本次话务服务的内容质量得分值为80分。 [0148] Emotion analysis submodule for recording user emotional analysis to determine user emotion parameter corresponding to a value; for example: the ratio of the total vocabulary of its service operator keywords contained in the recording was 18%, the proportion of the service keyword setting section in 80% 15% ~20 minutes, the contents of this traffic service quality score is 80 points.

[0149] 质量确定模块,用于根据上述得到的服务时长参数值、内容质量参数值和用户情感参数值,确定本次话务服务的服务质量测评值。 [0149] Quality determination module, according to the above-described length parameter values ​​obtained service quality parameter value and the content of user emotion parameter value, determining the service quality evaluation value of the current service traffic. 例如设定的测评规则中服务时长参数、内容质量参数和用户情感参数的权重比例为I : 2 : 2,则本次话务服务的服务质量测评值为[0150] 以上所述,仅为本发明较佳的具体实施方式,但本发明的保护范围并不局限于此,任何熟悉本技术领域的技术人员在本发明揭露的技术范围内,可轻易想到的变化、替换或应用到其他类似的装置,都应涵盖在本发明的保护范围之内。 For example the weight ratio Quanshi evaluation rule setting parameters in long service, and content quality parameter is user emotion parameter I: 2: 2, the evaluation of the quality of the current service traffic services is the [0150] above are only the invention preferred specific embodiments, but the scope of the present invention is not limited thereto, any skilled in the art in the art within the scope of the invention disclosed can be easily thought of the changes, substitutions, or applied to other similar means shall fall within the scope of the present invention. 因此,本发明的保护范围应该以权利要求书的保护范围为准。 Accordingly, the scope of the present invention, the scope of the claims should prevail.

Claims (12)

1. 一种呼叫中心服务质量检测方法,其特征在于,包括: 获取呼叫中心的服务录首; 提取所述服务录音中的属性信息,所述属性信息包括下列信息中的至少两种:服务时长、用户录音的性别及情感和话务员录音的话务内容; 根据所述属性信息,及所述属性信息与质量检测的参数的对应关系,确定用于质量检测的参数值;所述确定用于质量检测的参数值包含下列确定过程中的至少两种: 确定所述服务录音的服务时长,根据设置的服务时长阈值区间和服务时长参数的对应关系,查找获取到的服务录音的服务时长所在的阈值区间,确定服务录音的服务时长参数值; 确定出所述服务录音中包含的话务员录音,根据所述话务员录音的话务内容以及预先设定话务内容与内容质量参数的对应关系,确定对应的内容质量参数值; 确定出所述服务录音中包含的用 1. A call center service quality inspection method, characterized by comprising: obtaining first service record of the call center; extracting the attribute information of the service record, the attribute information comprises at least two of the following information: a service length , sex and emotional content traffic and attendant recording recording user; the corresponding relationship between the attribute information and the attribute information quality parameter detected, determining a parameter value detected by mass; for determining the mass the detected parameter values ​​comprise at least two of the following determination procedure: duration determining said service recording service, according to the correspondence size parameter duration threshold value interval and the service when the service is provided, the length where the lookup acquired service record of service threshold interval length determining parameter values ​​of the service recording services; operator determined the recording comprises recording service, according to the traffic of the operator content and the correspondence between the recording preset traffic content and content quality parameter, determining a corresponding SUMMARY quality parameter value; determined by recording the service included 录音,根据所述用户录音所对应的录音用户的性别和情感,以及预先设置的不同性别的各种情感与用户情感参数的对应关系,确定对应的用户情感参数值; 根据所述确定出的参数值,确定本次话务服务的服务质量测评值。 Recording, according to the user record corresponding to the user's gender and emotional recording, and the correspondence between different emotions and sex of user emotion parameter set in advance, determines the value of the emotion parameters corresponding to the user; according to the determined parameter value, fixed the quality of service measurement value traffic services.
2.如权利要求I所述的方法,其特征在于,所述根据所述参数值,确定本次话务服务的服务质量测评值具体包括: 根据设置的所述参数值在服务质量测评值中各自所占的权重比例,计算本次话务服务的服务质量测评值。 The value of the quality of service parameter values ​​in the evaluation: 2. The method of claim I, wherein the parameter value according to said determined quality of service evaluation value of the current service traffic comprises the proportion of their share of the heavy weights, calculate service quality evaluation value of this traffic services.
3.如权利要求I所述的方法,其特征在于,确定所述话务员录音所对应的内容质量参数值的过程,具体包括: 将所述话务员录音转化为录音文本,并计算统计出的所述录音文本中包含的预先指定的服务关键词的数量占统计出的所述录音文本中的词汇总量的比例值; 根据预先设定的比例区间与内容质量参数的对应关系,查找所述比例值所在的比例区间,确定本次服务录音对应的内容质量参数值。 3. The method of claim I, wherein the process of determining the content quality parameter value corresponding to the operator the recording comprises: recording into the recording text operator, and calculates the statistics predetermined number of recording the specified service keyword contained in the text value of the proportion of the total statistics of words in the transcriptions; according to a preset correspondence relationship between the content ratio range of quality parameters, to find the ratio value interval ratio where the determined content quality parameter value corresponding to the current service recording.
4.如权利要求I所述的方法,其特征在于,确定所述用户录音所对应的用户情感参数值的过程,具体包括: 根据预先确定的性别概率阈值以及计算出的所述用户录音基音信号累积密度函数值,确定所述用户录音所对应的录音用户的性别; 根据录音用户的性别、计算出的用户录音的情感特征参数以及预先统计出的不同性别各种情感所对应的情感特征参数分布区间,确定该录音用户的情感; 根据确定出的用户录音的性别和情感,查找预先设定的不同性别的各种情感与用户情感参数的对应关系,确定该用户录音的用户情感参数值。 4. The method of claim I, wherein the process of determining the values ​​of the emotion parameters corresponding to the user record user comprises: according to a predetermined probability threshold gender of the user, and recording the calculated pitch signal cumulative density function values ​​to determine the sex of the user record corresponding to the user record; recording according to the user's gender, affective characteristic parameters calculated in advance and recording the user's gender statistics emotions corresponding to the emotional parameters of the distribution interval, recording the user's emotion is determined; determined based on gender and emotional recording user, searches for the corresponding relation set in advance gender of emotions and user emotion parameter, determining user emotion parameter value of the user recording.
5.如权利要求2-4任一权项所述的方法,其特征在于,还包括:获取根据存储的话务员进行话务服务的操作信息,根据所述操作信息确定与所述服务录音对应的执行本次话务服务的话务员; 则所述确定本次话务服务的服务质量之后,还包括:将确定出的所述服务质量测评值以及执行本次话务服务的话务员进行存储。 Item 5. A method according to any one of claims 2-4 weight, characterized by further comprising: acquiring operation information for the traffic service according to the stored operator, according to the operation information corresponding to the determined recording service the operator performs traffic and services; after the determining the current quality of service traffic and services, further comprising: determining that the evaluation value of the quality of service and the operator for executing the stored traffic and services.
6.如权利要求5所述的方法,其特征在于,所述提取所述服务录音中的属性信息之前,还包括:通过小波变换消噪对获取的所述服务录音进行去噪处理。 6. The method according to claim 5, characterized in that, prior to attribute information of the service record of the extraction, further comprising: denoise the acquired service record by the wavelet transform denoising.
7. —种呼叫中心服务质量检测系统,其特征在于,包括: 录音获取模块,用于获取呼叫中心的服务录音; 提取模块,用于提取所述服务录音中的属性信息,所述属性信息包括下列信息中的至少两种:服务时长、用户录音的性别及情感和话务员录音的话务内容; 参数确定模块,用于根据所述属性信息,及所述属性信息与质量检测的参数的对应关系,确定用于质量检测的参数值;所述参数确定模块具体包括:时长分析子模块、确定子模块、质量分析子模块,以及情感分析子模块;所述时长分析子模块用于确定所述服务录音的服务时长,根据设置的服务时长阈值区间和服务时长参数的对应关系,查找获取到的服务录音的服务时长所在的阈值区间,确定服务录音的服务时长参数值;所述确定子模块,用于分别确定出所述录音获取模块获取到的服务录音中 7. - kind of call center services quality inspection system, characterized by comprising: recording an obtaining module, configured to obtain the call center service record; extraction means for extracting attribute information of the service record of the attribute information comprises at least two of the following information: a service length, sex and emotional content traffic and attendant recording recording user; parameter determination module, according to a corresponding relationship between the attribute information and the attribute information quality parameter detected determining a parameter value detected by mass; the parameter determining module comprises: a long analysis submodule, determining submodule, quality analysis sub-module, and emotional analysis sub-module; the duration of the analysis sub-module for determining a service long recording service, according to the correspondence size parameter duration threshold value interval and the service when the service is provided, the threshold interval length where the lookup acquired service record of the service, determine the length parameter values ​​services recording services; the determining sub-module, with were determined in the recording of the service acquisition module acquires recording 含的话务员录音和用户录音;所述质量分析子模块,用于根据所述话务员录音的话务内容以及预先设定话务内容与内容质量参数的对应关系,确定对应的内容质量参数值;所述情感分析子模块,用于根据所述用户录音所对应的录音用户的性别和情感,以及预先设置的不同性别的各种情感与用户情感参数的对应关系,确定对应的用户情感参数值; 质量确定模块,用于根据所述参数值,确定本次话务服务的服务质量。 Operator and the user recording comprising a recording; the quality analysis sub-module, according to the traffic of the operator content and the correspondence between the recording preset traffic content and content quality parameter is determined corresponding to the content quality parameter values; the said emotion analysis sub-module for recording corresponding relationship between the user corresponding to the user's gender and emotional recording, and a variety of user emotion and the emotion parameters set in advance gender, determining a parameter value corresponding to the user emotion; mass a determining module, according to the parameter value for determining the quality of service of the current service traffic.
8.如权利要求7所述的系统,其特征在于,所述质量分析子模块,具体包括: 转化单元,用于将所述话务员录音转化为录音文本; 计算单元,用于计算统计出的所述录音文本中包含的预先指定的服务关键词的数量占统计出的所述录音文本中的词汇总量的比例值; 质量确定单元,用于根据预先设定的比例区间与内容质量参数的对应关系,查找所述比例值所在的比例区间,确定本次服务录音对应的内容质量参数值。 8. The system according to claim 7, wherein said mass analysis sub-module comprises: conversion means for recording into the recording text operator; calculating unit for calculating statistics of the a pre-specified number of said recording service keyword contained in the text of the total vocabulary account statistics recording the text proportional value; quality determining unit configured to correspond to the content quality parameter according to the ratio of a preset interval relationship, to find the ratio of the interval where the ratio value to determine the contents of the current service quality parameter value corresponding to the recording.
9.如权利要求7所述的系统,其特征在于,所述情感分析子模块,具体包括: 性别确定单元,用于根据预先确定的性别概率阈值以及计算出的所述用户录音基音信号累积密度函数值,确定所述用户录音所对应的录音用户的性别; 情感确定单元,用于根据录音用户的性别、计算出的用户录音的情感特征参数以及预先统计出的不同性别各种情感所对应的情感特征参数分布区间,确定该录音用户的情感;参数确定单元,用于根据确定出的用户录音的性别和情感,查找预先设定的不同性别的各种情感与用户情感参数的对应关系,确定该用户录音的用户情感参数值。 9. The system according to claim 7, characterized in that the sentiment analysis sub-module comprises: means determining sex, gender for the cumulative density probability threshold a predetermined recording pitch signal and the user calculated function value, determines the sex of the user record corresponding to the user record; emotion determining means, corresponding to the recording according to the user's gender, affective characteristic parameters calculated from the user in advance and recording the statistics of different sexes emotions emotional characteristic parameter distribution section, recording the user's emotion is determined; parameter determining means for determining an emotion based on gender and recording of the user, searches for the corresponding relation set in advance gender of emotions and user emotion parameter, determined user emotion parameter value of the user recording.
10.如权利要求7所述的系统,其特征在于,所述质量确定模块,具体用于:根据设置的所述参数在服务质量测评值中各自所占的权重比例,计算本次话务服务的服务质量测评值。 10. The system according to claim 7, wherein said quality determination module is configured to: according to the weight ratio of the weight of the quality of service parameter set in the respective evaluation values ​​occupied by calculating the current service traffic assessment of quality of service value.
11.如权利要求7所述的系统,其特征在于,还包括: 消噪模块,用于通过小波变换消噪对获取的所述服务录音进行去噪处理。 11. The system according to claim 7, characterized in that, further comprising: noise canceling means for denoising the service record obtained by the wavelet transform denoising.
12.如权利要求7-11任一权项所述的系统,其特征在于,还包括: 信息处理模块,用于获取根据存储的话务员进行话务服务的操作信息,根据所述操作信息确定与所述服务录音对应的执行本次话务服务的话务员; 存储模块,用于将确定出的所述服务质量测评值以及执行本次话务服务的话务员进行存储。 Item 12. The system according to any of claims 7-11 a weight, characterized in that, further comprising: an information processing module for obtaining operation information storage service according to the traffic of the operator, and identification information according to the operation recording the service operator for executing the corresponding traffic and services; a storage module, for determining an evaluation value of the quality of service and the operator for executing the stored traffic and services.
CN 200910170363 2009-09-11 2009-09-11 Method and system for service quality detection for call center CN101662550B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 200910170363 CN101662550B (en) 2009-09-11 2009-09-11 Method and system for service quality detection for call center

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 200910170363 CN101662550B (en) 2009-09-11 2009-09-11 Method and system for service quality detection for call center

Publications (2)

Publication Number Publication Date
CN101662550A CN101662550A (en) 2010-03-03
CN101662550B true CN101662550B (en) 2012-10-03

Family

ID=41790321

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 200910170363 CN101662550B (en) 2009-09-11 2009-09-11 Method and system for service quality detection for call center

Country Status (1)

Country Link
CN (1) CN101662550B (en)

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103067621A (en) * 2011-10-18 2013-04-24 殷程 Home seat management system of call center and method thereof
CN103701999B (en) * 2012-09-27 2017-03-01 中国电信股份有限公司 The monitoring method of call center's voice call and system
CN103905657B (en) * 2012-12-28 2017-09-29 中国移动通信集团江苏有限公司 A kind of method and system of monitoring calling service quality
CN104158987B (en) * 2014-07-29 2017-03-29 北京佳讯飞鸿电气股份有限公司 A kind of method for lifting artificial telephone traffic system service quality
CN104680428B (en) * 2015-03-16 2018-03-16 朗新科技股份有限公司 A kind of construction method of power network CSAT model
CN106847305A (en) * 2015-12-04 2017-06-13 中国移动通信集团公司 A kind of method and device of the recording data for processing service calls
CN107093431A (en) * 2016-02-18 2017-08-25 中国移动通信集团辽宁有限公司 A kind of method and device that quality inspection is carried out to service quality
CN105975514A (en) * 2016-04-28 2016-09-28 朱宇光 Automatic quality testing method and system
CN106453971B (en) * 2016-11-22 2019-04-16 广东电网有限责任公司佛山供电局 The acquisition methods and call center's quality inspection system of call center's quality inspection voice
CN106776806A (en) * 2016-11-22 2017-05-31 广东电网有限责任公司佛山供电局 The methods of marking and system of call center's quality inspection voice
CN107154257A (en) * 2017-04-18 2017-09-12 苏州工业职业技术学院 Customer service quality evaluating method and system based on customer voice emotion
CN107293309A (en) * 2017-05-19 2017-10-24 四川新网银行股份有限公司 A kind of method that lifting public sentiment monitoring efficiency is analyzed based on customer anger

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101005540A (en) 2006-01-19 2007-07-25 北京讯鸟软件有限公司 Method for realizing sitting service level evaluation of calling center service
CN101150435A (en) 2007-10-24 2008-03-26 中兴通讯股份有限公司 A system and method for customer comment feedback by using multimedia

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101005540A (en) 2006-01-19 2007-07-25 北京讯鸟软件有限公司 Method for realizing sitting service level evaluation of calling center service
CN101150435A (en) 2007-10-24 2008-03-26 中兴通讯股份有限公司 A system and method for customer comment feedback by using multimedia

Also Published As

Publication number Publication date
CN101662550A (en) 2010-03-03

Similar Documents

Publication Publication Date Title
Saenz-Lechon et al. Methodological issues in the development of automatic systems for voice pathology detection
Henríquez et al. Characterization of healthy and pathological voice through measures based on nonlinear dynamics
US6638217B1 (en) Apparatus and methods for detecting emotions
US7725318B2 (en) System and method for improving the accuracy of audio searching
Wu et al. Automatic speech emotion recognition using modulation spectral features
Eskenazi et al. Acoustic correlates of vocal quality
US8078463B2 (en) Method and apparatus for speaker spotting
US7716048B2 (en) Method and apparatus for segmentation of audio interactions
CN1157710C (en) Speech datas extraction
US8412530B2 (en) Method and apparatus for detection of sentiment in automated transcriptions
WO2017084197A1 (en) Smart home control method and system based on emotion recognition
US8676586B2 (en) Method and apparatus for interaction or discourse analytics
US8219404B2 (en) Method and apparatus for recognizing a speaker in lawful interception systems
Yu et al. Audio classification from time-frequency texture
KR20140059754A (en) System and method for processing sound signals implementing a spectral motion transform
US8942356B2 (en) System and method for three-way call detection
BRPI0616903A2 (en) method for separating audio sources from a single audio signal, and, audio source classifier
US20140220526A1 (en) Customer sentiment analysis using recorded conversation
Lee et al. Modeling mutual influence of interlocutor emotion states in dyadic spoken interactions
US8005675B2 (en) Apparatus and method for audio analysis
Han et al. Acoustic classification of Australian anurans based on hybrid spectral-entropy approach
US10187525B2 (en) Communication session assessment
Stowell et al. Bird detection in audio: a survey and a challenge
Alexander et al. The effect of mismatched recording conditions on human and automatic speaker recognition in forensic applications
US20070061314A1 (en) Verbal web search with improved organization of documents based upon vocal gender analysis

Legal Events

Date Code Title Description
C06 Publication
C10 Request of examination as to substance
C14 Granted