Voiceprint recognition method and device based on distributed heterogeneous operation and storage medium

Abstract

The application relates to the field of artificial intelligence, and provides a voiceprint recognition method and device based on distributed heterogeneous operation, electronic equipment and a computer readable storage medium. The method comprises the following steps: obtaining a to-be-recognized voice signal; performing task splitting on a preset voiceprint recognition system to obtain a plurality of subtask modules; performing evaluation processing on the subtask modules based on a preset dense calculation evaluation model to obtain dense calculation type information; dividing the plurality of subtask modules into a preset CPU calculation-intensive microservice pool and a preset GPU calculation-intensive microservice pool based on the dense calculation type information; performing first calculation processing on the to-be-recognized voice signal based on the subtask modules in the CPU calculation-intensive microservice pool to obtain first processing information; and performing second calculation processing on the first processing information based on the subtask modules in the GPU calculation-intensive microservice pool to obtain a voiceprint recognition result. The above technical solution can improve the efficiency of voiceprint recognition.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to, but are not limited to, the field of artificial intelligence, and in particular to a voiceprint recognition method and device based on distributed heterogeneous operation, an electronic device, and a computer readable storage medium. BACKGROUND

[0002] In the field of voiceprint recognition, in order to obtain a voiceprint recognition result, speech feature extraction is usually performed first, and then the extracted speech features are input into a voiceprint recognition model for calculation to obtain voiceprint features. However, in the process of speech feature extraction and voiceprint feature recognition, some of the computing units used are central processing unit (CPU) computation-intensive, and some are graphics processing unit (GPU) computation-intensive. These two types of computation are cross-overlapped in the voiceprint recognition system, resulting in a low queries per second (QPS) of the voiceprint recognition system, which in turn affects the efficiency of voiceprint recognition. SUMMARY

[0003] The following is an overview of the subject matter of the detailed description herein. This overview is not intended to limit the scope of the claims.

[0004] To solve the problems mentioned in the background, the embodiments of the present application provide a voiceprint recognition method and device based on distributed heterogeneous operation, an electronic device, and a computer readable storage medium, which can improve the efficiency of voiceprint recognition.

[0005] In a first aspect, the embodiments of the present application provide a voiceprint recognition method based on distributed heterogeneous operation, which comprises:

[0006] obtaining a to-be-recognized speech signal;

[0007] task splitting is performed on a preset voiceprint recognition system to obtain a plurality of subtask modules;

[0008] Each subtask module is evaluated based on a preset dense computing evaluation model to obtain corresponding dense computing type information of the subtask module;

[0009] Based on the dense computing type information corresponding to each subtask module, the plurality of subtask modules are divided into a preset central processing unit (CPU) computation-intensive microservice pool and a preset graphics processing unit (GPU) computation-intensive microservice pool;

[0010] The to-be-recognized speech signal is subjected to first computing processing based on the subtask modules of the CPU computation-intensive microservice pool to obtain first processing information;

[0011] The first processing information is subjected to second calculation processing by a subtask module of the GPU calculation-intensive microservice pool, and a voiceprint recognition result is obtained.

[0012] According to the voiceprint recognition method based on distributed heterogeneous operation provided in the embodiments of the present application, the following beneficial effects are achieved. Firstly, a to-be-recognized voice signal is acquired. Then, a preset voiceprint recognition system is subjected to task splitting, and a plurality of subtask modules are obtained. Then, each subtask module is subjected to evaluation processing based on a preset dense calculation evaluation model, and dense calculation type information corresponding to each subtask module is obtained. Then, the plurality of subtask modules are divided into a preset CPU calculation-intensive microservice pool and a preset GPU calculation-intensive microservice pool based on the dense calculation type information corresponding to each subtask module. Then, the to-be-recognized voice signal is subjected to first calculation processing by a subtask module of the CPU calculation-intensive microservice pool, and first processing information is obtained. Finally, the first processing information is subjected to second calculation processing by a subtask module of the GPU calculation-intensive microservice pool, and a voiceprint recognition result is obtained. Through the above technical solution, the CPU calculation-intensive part of voiceprint recognition is placed on a CPU machine, and the GPU calculation-intensive part is placed on a GPU machine for acceleration, thereby avoiding the problem that the CPU is overloaded and the GPU is idle in the traditional voiceprint recognition process, achieving acceleration of the overall system, and improving the efficiency of voiceprint recognition.

[0013] According to some embodiments of the present application, the evaluation processing of each subtask module based on the preset dense calculation evaluation model to obtain the dense calculation type information corresponding to the subtask module comprises the following steps.

[0014] The interface of each subtask module is subjected to unified processing and microservice reconstruction, and a plurality of microservice modules with unified interfaces are obtained.

[0015] Each microservice module is subjected to third calculation processing based on the dense calculation evaluation model, and judgment data corresponding to each microservice module is obtained.

[0016] The judgment data is subjected to fourth calculation processing based on a preset dense calculation judgment algorithm, and the dense calculation type information corresponding to the subtask module is obtained.

[0017] According to some embodiments of the present application, the division of the plurality of subtask modules into the preset CPU calculation-intensive microservice pool and the preset GPU calculation-intensive microservice pool based on the dense calculation type information corresponding to each subtask module comprises the following steps.

[0018] A dense calculation type discrimination parameter is determined according to the dense calculation type information.

[0019] According to the dense calculation type discrimination parameter and a preset dense calculation type discrimination threshold, a dense calculation type discrimination result is obtained.

[0020] According to the dense calculation type discrimination result, the plurality of sub-task modules are divided into a preset central processing unit (CPU) computation-intensive micro-service pool and a preset graphics processing unit (GPU) computation-intensive micro-service pool.

[0021] According to some embodiments of the present application, the dense calculation type discrimination result comprises a first discrimination result and a second discrimination result, and the obtaining of the dense calculation type discrimination result according to the dense calculation type discrimination parameter and the preset dense calculation type discrimination threshold comprises:

[0022] performing a difference operation on the dense calculation type discrimination parameter and the dense calculation type discrimination threshold;

[0023] in a case where the dense calculation type discrimination parameter is greater than the dense calculation type discrimination threshold, the first discrimination result is obtained;

[0024] in a case where the dense calculation type discrimination parameter is not greater than the dense calculation type discrimination threshold, the second discrimination result is obtained.

[0025] According to some embodiments of the present application, the dividing of the plurality of sub-task modules into the preset CPU computation-intensive micro-service pool and the preset GPU computation-intensive micro-service pool according to the dense calculation type discrimination result comprises:

[0026] in a case where the dense calculation type discrimination result is the first discrimination result, the corresponding sub-task module is divided into the CPU computation-intensive micro-service pool;

[0027] in a case where the dense calculation type discrimination result is the second discrimination result, the corresponding sub-task module is divided into the GPU computation-intensive micro-service pool.

[0028] According to some embodiments of the present application, the judgment data comprises CPU usage, CPU weight, memory usage, memory weight, interface load, interface weight, GPU usage, GPU weight, video memory usage and video memory weight, and the dense calculation judgment algorithm formula is as follows:

[0029]

[0030] Wherein, K represents the dense calculation type information, a represents the CPU usage rate, x represents the CPU weight, b represents the memory usage rate, y represents the memory weight, c represents the interface load, z represents the interface weight, d represents the GPU usage rate, f represents the GPU weight, e represents the video memory usage rate, and g represents the video memory weight.

[0031] According to some embodiments of the present application, the preset voiceprint recognition system is task-splitting to obtain a plurality of sub-task modules, which includes:

[0032] According to the preset voiceprint recognition task, the corresponding sub-task module is extracted from the voiceprint recognition system.

[0033] The sub-task module includes a format conversion module, a channel separation module, a noise removal module, a mel frequency cepstrum coefficient extraction module, a frequency domain feature extraction module, a cepstrum mean normalization module, a speech endpoint detection module, and a voiceprint feature extraction module.

[0034] In a second aspect, the embodiments of the present application also provide a voiceprint recognition device based on distributed heterogeneous operation, which includes:

[0035] The first processing module is configured to acquire a to-be-recognized voice signal.

[0036] The second processing module is configured to perform task-splitting on a preset voiceprint recognition system to obtain a plurality of sub-task modules.

[0037] The third processing module is configured to perform evaluation processing on each sub-task module based on a preset dense calculation evaluation model to obtain the dense calculation type information corresponding to the sub-task module.

[0038] The fourth processing module is configured to divide the plurality of sub-task modules into a preset central processing unit (CPU) calculation-intensive micro-service pool and a preset graphics processing unit (GPU) calculation-intensive micro-service pool based on the dense calculation type information corresponding to each sub-task module.

[0039] The fifth processing module is configured to perform first calculation processing on the to-be-recognized voice signal based on the sub-task module of the CPU calculation-intensive micro-service pool to obtain first processing information.

[0040] The sixth processing module is configured to perform second calculation processing on the first processing information based on the sub-task module of the GPU calculation-intensive micro-service pool to obtain a voiceprint recognition result.

[0041] In a third aspect, an electronic device is provided, which includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor, and the processor implements the voiceprint recognition method based on distributed heterogeneous computing according to the first aspect when executing the computer program.

[0042] In a fourth aspect, a computer readable storage medium is provided, which stores computer executable instructions for executing the voiceprint recognition method based on distributed heterogeneous computing according to the first aspect.

[0043] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent from the description, or can be learned by practice of the present application. The objects and other advantages of the present application will be realized and attained by the structure particularly pointed out in the written description and claims thereof as well as the appended drawings. BRIEF DESCRIPTION OF DRAWINGS

[0044] The accompanying drawings are included to provide a further understanding of the technical solutions of the present application, and constitute a part of the specification, and are used to explain the technical solutions of the present application together with the embodiments of the present application, and do not constitute a limitation to the technical solutions of the present application.

[0045] Figure 1 is a flowchart of the voiceprint recognition method based on distributed heterogeneous computing provided by an embodiment of the present application;

[0046] Figure 2 is a flowchart of the evaluation processing in the voiceprint recognition method based on distributed heterogeneous computing provided by an embodiment of the present application;

[0047] Figure 3 is a flowchart of the division of the subtask module in the voiceprint recognition method based on distributed heterogeneous computing provided by an embodiment of the present application;

[0048] Figure 4 is a flowchart of the acquisition of the judgment result in the voiceprint recognition method based on distributed heterogeneous computing provided by an embodiment of the present application;

[0049] Figure 5 is a flowchart of the division of the subtask module in the voiceprint recognition method based on distributed heterogeneous computing provided by another embodiment of the present application;

[0050] Figure 6 is a flowchart of the voiceprint recognition method based on distributed heterogeneous computing provided by another embodiment of the present application;

[0051] Figure 7 is a schematic diagram of the voiceprint recognition device based on distributed heterogeneous computing provided by an embodiment of the present application;

[0052] Figure 8 This is a schematic diagram of an electronic device provided in one embodiment of this application. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0054] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0055] It should be noted that, unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0056] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0057] AI is a new technical science that studies and develops theories, methods, technologies, and application systems for simulating, extending, and expanding human intelligence. Artificial intelligence is a branch of computer science that attempts to understand the essence of intelligence and produce new intelligent machines that can react in a way similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems. Artificial intelligence can simulate the information processes of human consciousness and thought. Furthermore, artificial intelligence utilizes digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceiving the environment, acquiring knowledge, and using that knowledge to achieve optimal results—the theories, methods, technologies, and application systems available for use.

[0058] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0059] Artificial intelligence, or AI, is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0060] The servers involved in artificial intelligence technology can be standalone servers or cloud servers that provide basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0061] This application provides a voiceprint recognition method, device, electronic device, and computer-readable storage medium based on distributed heterogeneous computing. First, the voice signal to be recognized is acquired. Then, a pre-defined voiceprint recognition system is task-split into multiple sub-task modules. Next, each sub-task module is evaluated based on a pre-defined intensive computing evaluation model to obtain intensive computing type information corresponding to the sub-task module. Then, based on the intensive computing type information corresponding to each sub-task module, the multiple sub-task modules are divided into a pre-defined CPU-intensive microservice pool and a pre-defined GPU-intensive microservice pool. Next, the sub-task modules in the CPU-intensive microservice pool perform a first computational processing on the voice signal to be recognized to obtain first processing information. Finally, the sub-task modules in the GPU-intensive microservice pool perform a second computational processing on the first processing information to obtain the voiceprint recognition result. Through the above technical solution, the CPU-intensive part of voiceprint recognition is placed on a CPU machine, and the GPU-intensive part is placed on a GPU machine for acceleration, avoiding the problem of CPU overload and GPU idleness in the traditional voiceprint recognition process, thus accelerating the overall system and improving the efficiency of voiceprint recognition.

[0062] The voiceprint recognition method based on distributed heterogeneous computing provided in this application relates to the field of artificial intelligence technology. This method can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application implementing the voiceprint recognition method based on distributed heterogeneous computing, but is not limited to the above forms.

[0063] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0064] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards of the relevant countries and regions. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirects to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data for the proper functioning of the embodiments of this application obtained.

[0065] The embodiments of this application will be further described below with reference to the accompanying drawings.

[0066] like Figure 1 As shown,Figure 1 This is a flowchart of a voiceprint recognition method based on distributed heterogeneous computing provided in one embodiment of this application. The voiceprint recognition method based on distributed heterogeneous computing includes, but is not limited to, steps S100 to S600.

[0067] Step S100: Acquire the speech signal to be recognized;

[0068] Step S200: The preset voiceprint recognition system is divided into multiple sub-task modules.

[0069] Step S300: Evaluate each sub-task module based on the preset intensive computing evaluation model to obtain the intensive computing type information corresponding to the sub-task module.

[0070] Step S400: Based on the intensive computing type information corresponding to each sub-task module, the multiple sub-task modules are divided into a preset CPU computing-intensive microservice pool and a preset GPU computing-intensive microservice pool.

[0071] Step S500: The sub-task module based on the CPU-intensive microservice pool performs the first computational processing on the speech signal to be recognized to obtain the first processing information.

[0072] In step S600, the subtask module based on the GPU computing-intensive microservice pool performs a second computational process on the first processing information to obtain the voiceprint recognition result.

[0073] It should be noted that the process involves several steps: First, the voice signal to be recognized is acquired. Then, the preset voiceprint recognition system is task-splitting to obtain multiple sub-task modules. Next, each sub-task module is evaluated based on a preset intensive computation evaluation model, yielding intensive computation type information for each sub-task module. Then, based on the intensive computation type information for each sub-task module, the multiple sub-task modules are divided into a preset CPU-intensive microservice pool and a preset GPU-intensive microservice pool. Next, the sub-task modules in the CPU-intensive microservice pool perform a first computational processing on the voice signal to be recognized, obtaining first processing information. Finally, the sub-task modules in the GPU-intensive microservice pool perform a second computational processing on the first processing information to obtain the voiceprint recognition result. This technical solution accelerates the voiceprint recognition process by placing the CPU-intensive part of the voiceprint recognition process on a CPU machine and the GPU-intensive part on a GPU machine, avoiding the problem of CPU overload and GPU idleness in traditional voiceprint recognition processes. This achieves overall system acceleration and improves the efficiency of voiceprint recognition.

[0074] It is worth noting that the voice signal to be recognized is the voice signal that needs to be recognized and processed. The voice signal can be the sound signal emitted by the person being tested on site, or the sound signal collected by the voice collection device. There is no limitation here.

[0075] It should be noted that the voice signals to be recognized acquired in the embodiments of the present invention will be subject to the user's permission or consent beforehand. Furthermore, the collection, use, and processing of this data will comply with the relevant laws, regulations, and standards of the relevant countries and regions. In addition, when the embodiments of this application need to obtain the user's sensitive personal information, the user's separate permission or consent will be obtained through pop-ups or redirection to a confirmation page. Only after obtaining the user's separate permission or consent will the necessary user-related data for the normal operation of the embodiments of this application be acquired.

[0076] It is understandable that CPU-intensive computing means that a large number of calculations rely on the CPU; while GPU-intensive computing means that a large number of calculations rely on the GPU.

[0077] It should be noted that traditional voiceprint recognition systems include multiple sub-task modules. Some sub-task modules are mainly used for speech feature extraction, while others are mainly used for voiceprint feature extraction. Some sub-task modules are CPU-intensive, while others are GPU-intensive.

[0078] Understandably, the sub-task module based on the CPU-intensive microservice pool performs the first computational processing on the speech signal to be recognized, obtaining the first processed information; then, the sub-task module based on the GPU-intensive microservice pool performs the second computational processing on the first processed information, obtaining the voiceprint recognition result. This can greatly improve the overall network inference throughput, placing the CPU-intensive part of voiceprint recognition on the CPU machine and the GPU-intensive part on the GPU machine for acceleration, avoiding the problem of CPU overload and GPU idleness in the traditional voiceprint recognition process, thus accelerating the overall system. It also improves the utilization of the graphics card, avoiding the problem of GPU idleness caused by CPU overload, thus greatly improving the utilization of the GPU and effectively utilizing valuable graphics card resources.

[0079] It's important to note that the CPU, as the core of a computer system's computation and control, is the final execution unit for information processing and program execution. The CPU emerged in the era of large-scale integrated circuits, and its continuous development and improvement have been driven by iterative updates in processor architecture design and advancements in integrated circuit technology. From its initial focus on mathematical calculations to its widespread application in general-purpose computing, it has evolved from 4-bit to 8-bit, 16-bit, 32-bit processors, and finally to 64-bit processors. The CPU is one of the main components of an electronic computer, a core part of the computer. Its primary function is to interpret computer instructions and process data in computer software. The CPU is the core component responsible for reading, decoding, and executing instructions. The central processing unit mainly consists of two parts: the control unit and the arithmetic logic unit (ALU), which also includes high-speed cache memory and the data and control buses that connect them. The three core components of an electronic computer are the CPU, internal memory, and input / output devices. The central processing unit's main functions are processing instructions, executing operations, controlling timing, and processing data. In computer architecture, the CPU is the core hardware unit that controls and allocates all the computer's hardware resources (such as memory and input / output units) and performs general-purpose calculations. The CPU is the core of the computer's computation and control. All software layer operations in a computer system are ultimately mapped to CPU operations through the instruction set.

[0080] It's important to note that a GPU, also known as a graphics processing unit (GPU), is a microprocessor specifically designed for image and graphics processing in personal computers, workstations, game consoles, and some mobile devices (such as tablets and smartphones). A raster display system cannot function without a GPU; the GPU is a crucial component of the graphics system architecture and acts as the link between the computer and the display terminal. Essentially, every display system has a GPU (commonly known as a graphics card). However, early graphics cards only contained simple memory and buffers, essentially serving only the function of storing and transferring graphics; all operations had to be controlled by the CPU. This was sufficient for text and some simple graphics, but when dealing with complex scenes, especially realistic 3D scenes, this system alone was insufficient. Therefore, later graphics cards incorporated graphics processing capabilities. They not only stored graphics but also performed most graphics functions, significantly reducing the CPU's workload and improving display capabilities and speed.

[0081] In some embodiments, such as Figure 2 As shown, step S300 may include, but is not limited to, steps S310 to S330.

[0082] Step S310: Perform unified interface processing and microservice transformation on each subtask module to obtain multiple microservice modules with unified interfaces.

[0083] Step S320: Perform a third computational process on each microservice module based on the intensive computational evaluation model to obtain the judgment data corresponding to each microservice module;

[0084] Step S330: Based on the preset dense computing judgment algorithm, the judgment data is processed by a fourth calculation to obtain the dense computing type information corresponding to the subtask module.

[0085] It should be noted that in the process of evaluating sub-task modules, each sub-task module is first subjected to unified interface processing and microservice transformation to obtain multiple microservice modules with unified interfaces; then, a third computational processing is performed on each microservice module based on the intensive computation evaluation model to obtain the judgment data corresponding to each microservice module; finally, a fourth computational processing is performed on the judgment data based on the preset intensive computation judgment algorithm to obtain the intensive computation type information corresponding to the sub-task module.

[0086] It is worth noting that each sub-task module undergoes unified interface processing and microservice transformation to prepare for subsequent input into the intensive computing evaluation model.

[0087] It should be noted that the intensive computing evaluation model includes standard test data (voice), execution scripts, and result analysis scripts. Each microservice module is input into the intensive computing evaluation model and run once to obtain the judgment data for each microservice module.

[0088] In some embodiments, such as Figure 3 As shown, step S400 may include, but is not limited to, steps S410 to S430.

[0089] Step S410: Determine the dense computing type discrimination parameters based on the dense computing type information;

[0090] Step S420: Obtain the dense computing type discrimination result based on the dense computing type discrimination parameters and the preset dense computing type discrimination threshold;

[0091] Step S430: Based on the intensive computing type discrimination result, the multiple sub-task modules are divided into the preset CPU computing intensive microservice pool and the preset GPU computing intensive microservice pool.

[0092] It should be noted that in the process of dividing the subtask modules, the dense computing type discrimination parameters are first determined based on the dense computing type information; then the dense computing type discrimination parameters are compared with the preset dense computing type discrimination threshold to obtain the dense computing type discrimination result; finally, based on the dense computing type discrimination result, multiple subtask modules are divided into CPU computing-intensive microservice pools and GPU computing-intensive microservice pools.

[0093] It's worth noting that after dividing multiple sub-task modules into CPU-intensive and GPU-intensive microservice pools, the size of the microservice pools and the number of certain sub-task modules can be dynamically adjusted based on actual business needs. For example, in batch format conversion, since feature extraction is not required, the number of format conversion microservices can be increased while the number of feature extraction microservices can be reduced, thereby improving the utilization efficiency of computing hardware.

[0094] In some embodiments, such as Figure 4 As shown, the dense computing type discrimination result includes a first discrimination result and a second discrimination result. The above step S420 may include, but is not limited to, steps S421 to S423.

[0095] Step S421: Perform a difference operation between the dense computing type discrimination parameter and the dense computing type discrimination threshold;

[0096] Step S422: If the dense computing type discrimination parameter is greater than the dense computing type discrimination threshold, obtain the first discrimination result;

[0097] Step S423: If the dense computation type discrimination parameter is not greater than the dense computation type discrimination threshold, a second discrimination result is obtained.

[0098] It should be noted that in the process of judging the subtask module, the difference between the dense computing type judgment parameter and the dense computing type judgment threshold is first calculated; then, if the dense computing type judgment parameter is greater than the dense computing type judgment threshold, the first judgment result will be obtained; and if the dense computing type judgment parameter is not greater than the dense computing type judgment threshold, the second judgment result will be obtained.

[0099] In some embodiments, such as Figure 5 As shown, step S430 may include, but is not limited to, steps S431 to S432.

[0100] Step S431: If the dense computing type discrimination result is the first discrimination result, the corresponding subtask module is assigned to the CPU computing intensive microservice pool.

[0101] Step S432: If the dense computing type discrimination result is the second discrimination result, the corresponding subtask module is assigned to the GPU computing-intensive microservice pool.

[0102] It should be noted that during the partitioning process, when the intensive computing type is determined to be the first result, the corresponding sub-task module will be partitioned into the CPU intensive microservice pool; while when the intensive computing type is determined to be the second result, the corresponding sub-task module will be partitioned into the GPU intensive microservice pool.

[0103] It's important to note that microservices, also known as microservice architecture, are a software architecture approach. It builds an application into a series of small, autonomous services divided into modules based on business domains. In a microservice architecture, each service is self-contained and implements a single business function. A microservice pool is a collection of multiple microservices that together perform a specific function.

[0104] In some embodiments, the judgment data includes CPU utilization, CPU weight, memory utilization, memory weight, interface load, interface weight, GPU utilization, GPU weight, video memory utilization, and video memory weight. The formula for the intensive calculation judgment algorithm is as follows:

[0105]

[0106] Where K represents intensive computing type information, a represents CPU utilization, x represents CPU weight, b represents memory utilization, y represents memory weight, c represents interface load, z represents interface weight, d represents GPU utilization, f represents GPU weight, e represents video memory utilization, and g represents video memory weight.

[0107] It should be noted that the intensive computing type information is obtained through the above formula, and then the intensive computing type discrimination parameters are determined based on the intensive computing type information. Next, the intensive computing type discrimination parameters are compared with the preset intensive computing type discrimination threshold to obtain the intensive computing type discrimination result. Finally, based on the intensive computing type discrimination result, multiple sub-task modules are divided into CPU computing intensive microservice pools and GPU computing intensive microservice pools.

[0108] In some embodiments, such as Figure 6 As shown, step S200 may include, but is not limited to, step S210.

[0109] Step S210: Extract the corresponding sub-task module from the voiceprint recognition system according to the preset voiceprint recognition task.

[0110] The sub-task modules include: format conversion module, channel separation module, noise reduction module, Mel frequency cepstral coefficient extraction module, frequency domain feature extraction module, cepstral mean normalization module, speech endpoint detection module, and voiceprint feature extraction module.

[0111] It should be noted that, based on the preset voiceprint recognition task, the corresponding sub-task modules are extracted from the voiceprint recognition system; among them, the sub-task modules may include a format conversion module, a vocal tract separation module, a noise reduction module, a Mel frequency cepstral coefficient extraction module, a frequency domain feature extraction module, a cepstral mean normalization module, a speech endpoint detection module, and a voiceprint feature extraction module.

[0112] It is worth noting that the format conversion module, the channel separation module, the noise reduction module, the Mel frequency cepstral coefficient extraction module, the frequency domain feature extraction module, the cepstral mean normalization module, and the speech endpoint detection module are all CPU-intensive sub-task modules; while the voiceprint feature extraction module is a GPU-intensive sub-task module.

[0113] The above technical solution first acquires the speech signal to be recognized; then, the preset voiceprint recognition system is task-splitting to obtain multiple sub-task modules; next, each sub-task module is evaluated based on a preset intensive computing evaluation model to obtain the intensive computing type information corresponding to the sub-task module; then, based on the intensive computing type information corresponding to each sub-task module, the multiple sub-task modules are divided into a preset CPU-intensive microservice pool and a preset GPU-intensive microservice pool; next, the sub-task modules based on the CPU-intensive microservice pool perform a first computational processing on the speech signal to be recognized to obtain first processing information; finally, the sub-task modules based on the GPU-intensive microservice pool perform a second computational processing on the first processing information to obtain the voiceprint recognition result. This technical solution accelerates the voiceprint recognition process by placing the CPU-intensive part of the CPU-intensive part on the CPU machine and the GPU-intensive part on the GPU machine, avoiding the problem of CPU overload and GPU idleness in traditional voiceprint recognition processes, thus accelerating the overall system and improving the efficiency of voiceprint recognition.

[0114] In addition, such as Figure 7 As shown, one embodiment of this application also provides a voiceprint recognition device 10 based on distributed heterogeneous computing, comprising:

[0115] The first processing module 100 is used to acquire the speech signal to be recognized;

[0116] The second processing module 200 is used to split the preset voiceprint recognition system into multiple sub-task modules.

[0117] The third processing module 300 is used to evaluate each sub-task module based on a preset intensive computing evaluation model to obtain the intensive computing type information corresponding to the sub-task module.

[0118] The fourth processing module 400 is used to divide multiple sub-task modules into a preset CPU computing-intensive microservice pool and a preset GPU computing-intensive microservice pool based on the intensive computing type information corresponding to each sub-task module.

[0119] The fifth processing module 500 is used to perform the first computational processing on the speech signal to be recognized by the sub-task module based on the CPU computationally intensive microservice pool to obtain the first processing information.

[0120] The sixth processing module 600 is used by the sub-task module based on the GPU computing-intensive microservice pool to perform a second computational process on the first processing information to obtain the voiceprint recognition result.

[0121] In one embodiment, the voice signal to be recognized is first acquired; then, the preset voiceprint recognition system is task-splitting to obtain multiple sub-task modules; then, each sub-task module is evaluated based on a preset intensive computing evaluation model to obtain the intensive computing type information corresponding to the sub-task module; next, based on the intensive computing type information corresponding to each sub-task module, the multiple sub-task modules are divided into a preset CPU-intensive microservice pool and a preset GPU-intensive microservice pool; then, the sub-task modules based on the CPU-intensive microservice pool perform a first computational processing on the voice signal to be recognized to obtain first processing information; finally, the sub-task modules based on the GPU-intensive microservice pool perform a second computational processing on the first processing information to obtain the voiceprint recognition result; through the above technical solution, the CPU-intensive part of voiceprint recognition is placed on the CPU machine, and the GPU-intensive part is placed on the GPU machine for acceleration, avoiding the problem of CPU overload and GPU idleness in the traditional voiceprint recognition process, realizing the acceleration of the overall system and improving the efficiency of voiceprint recognition.

[0122] The specific implementation of the voiceprint recognition device based on distributed heterogeneous computing is basically the same as the specific implementation of the voiceprint recognition method based on distributed heterogeneous computing described above, and will not be repeated here.

[0123] In addition, such as Figure 8 As shown, one embodiment of this application also provides an electronic device 700, which includes: a memory 710, a processor 720, and a computer program stored on the memory 710 and executable on the processor 720.

[0124] The processor 720 and memory 710 can be connected via a bus or other means.

[0125] The non-transient software program and instructions required to implement the voiceprint recognition method based on distributed heterogeneous computing in the above embodiments are stored in memory 710. When executed by processor 720, the voiceprint recognition method based on distributed heterogeneous computing in the above embodiments is executed, for example, the method described above is executed. Figure 1 Method steps S100 to S600 in the text Figure 2 Method steps S310 to S330 in the text Figure 3 Method steps S410 to S430 in the text Figure 4 Method steps S421 to S423 in the text Figure 5 Method steps S431 to S432 in the text Figure 6 Method steps S100 to S600.

[0126] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0127] Furthermore, one embodiment of this application also provides a computer-readable storage medium storing computer-executable instructions that are executed by a processor 720 or a controller, for example, by a processor 720 in the above-described device embodiment. These instructions enable the processor 720 to execute the voiceprint recognition method based on distributed heterogeneous computing described above, for example, to execute the above-described... Figure 1 Method steps S100 to S600 in the text Figure 2 Method steps S310 to S330 in the text Figure 3 Method steps S410 to S430 in the text Figure 4 Method steps S421 to S423 in the text Figure 5 Method steps S431 to S432 in the text Figure 6 Method steps S100 to S600.

[0128] The above embodiments can be used in combination, and modules with the same name in different embodiments may be the same or different.

[0129] The foregoing has described specific embodiments of this application; other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims may be performed in a different order than those shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily have to follow the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0130] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and computer-readable storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0131] The apparatus, device, computer-readable storage medium and method provided in the embodiments of this application are corresponding. Therefore, the apparatus, device and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding apparatus, device and computer storage medium will not be described again here.

[0132] In the 1990s, improvements to a technology could be clearly distinguished as either hardware improvements (e.g., improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the methodology). However, with technological advancements, many methodological improvements today can be considered direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved methodology into the hardware circuit. Therefore, it cannot be said that a methodological improvement cannot be implemented using hardware physical modules. For example, a Programmable Logic Device (PLD) (such as a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logic function is determined by the user programming the device. Designers can program and "integrate" a digital system onto a PLD themselves, without needing chip manufacturers to design and manufacture dedicated integrated circuit chips. Furthermore, nowadays, instead of manually manufacturing integrated circuit chips, this programming is mostly implemented using "logic compiler" software. Similar to the software compiler used in program development, the original code before compilation must be written in a specific programming language, called a Hardware Description Language (HDL). There are many HDLs, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, and RHDL (Ruby Hardware Description Language). Currently, the most commonly used are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should understand that by simply performing some logic programming on the method flow using one of these hardware description languages ​​and programming it into an integrated circuit, the hardware circuit implementing the logical method flow can be easily obtained.

[0133] The controller can be implemented in any suitable manner. For example, it can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code form, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included therein for implementing various functions can also be considered as structures within the hardware component. Alternatively, the means for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0134] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0135] For ease of description, the above apparatus is described by dividing it into various functional units. Of course, in implementing the embodiments of this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0136] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, embodiments of this application can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, embodiments of this application can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0137] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. 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 apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0138] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0139] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0140] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0141] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0142] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0143] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0144] In this application embodiment, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent the existence of A alone, A and B simultaneously, or B alone. A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of singular or plural items. For example, at least one of a, b, and c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c can be single or multiple.

[0145] The embodiments of this application can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. The embodiments of this application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In a distributed computing environment, program modules can reside in local and remote computer storage media, including storage devices.

[0146] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0147] The above description is merely an embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of this application should be included within the scope of the claims of this application.

Claims

1. A voiceprint recognition method based on distributed heterogeneous computing, characterized in that, The method includes: Acquire the speech signal to be recognized; The preset voiceprint recognition system is divided into multiple sub-task modules. Each sub-task module is evaluated based on a preset intensive computing evaluation model to obtain the intensive computing type information corresponding to the sub-task module. Based on the intensive computing type information corresponding to each of the sub-task modules, the multiple sub-task modules are divided into a preset CPU computing-intensive microservice pool and a preset GPU computing-intensive microservice pool. The subtask module of the CPU-intensive microservice pool performs a first calculation on the speech signal to be recognized to obtain first processing information. The subtask module of the GPU-intensive microservice pool performs a second computational process on the first processed information to obtain the voiceprint recognition result. The evaluation process for each sub-task module based on a preset intensive computing evaluation model, to obtain intensive computing type information corresponding to the sub-task module, includes: Each sub-task module undergoes unified interface processing and microservice transformation to obtain multiple microservice modules with unified interfaces. Based on the intensive computing evaluation model, a third computational process is performed on each microservice module to obtain judgment data corresponding to each microservice module; wherein, the judgment data includes CPU utilization, CPU weight, memory utilization, memory weight, interface load, interface weight, GPU utilization, GPU weight, video memory utilization, and video memory weight. Based on a preset intensive computing judgment algorithm, a fourth calculation process is performed on the CPU utilization rate, CPU weight, memory utilization rate, memory weight, interface load, interface weight, GPU utilization rate, GPU weight, video memory utilization rate, and video memory weight to obtain the intensive computing type information corresponding to the subtask module.

2. The voiceprint recognition method based on distributed heterogeneous computing according to claim 1, characterized in that, The step of dividing the multiple sub-task modules into a preset CPU-intensive microservice pool and a preset GPU-intensive microservice pool based on the computationally intensive type information corresponding to each sub-task module includes: Determine the dense computing type discrimination parameters based on the dense computing type information; Based on the dense computing type discrimination parameters and the preset dense computing type discrimination threshold, the dense computing type discrimination result is obtained; Based on the intensive computing type discrimination result, the multiple sub-task modules are divided into a preset CPU computing-intensive microservice pool and a preset GPU computing-intensive microservice pool.

3. The voiceprint recognition method based on distributed heterogeneous computing according to claim 2, characterized in that, The dense computing type discrimination result includes a first discrimination result and a second discrimination result. The step of obtaining the dense computing type discrimination result based on the dense computing type discrimination parameters and a preset dense computing type discrimination threshold includes: The difference between the dense computing type discrimination parameter and the dense computing type discrimination threshold is calculated. If the dense computing type discrimination parameter is greater than the dense computing type discrimination threshold, the first discrimination result is obtained; If the dense computation type discrimination parameter is not greater than the dense computation type discrimination threshold, the second discrimination result is obtained.

4. The voiceprint recognition method based on distributed heterogeneous computing according to claim 3, characterized in that, The step of dividing the multiple sub-task modules into a preset CPU-intensive microservice pool and a preset GPU-intensive microservice pool based on the intensive computing type determination result includes: If the intensive computing type determination result is the first determination result, the corresponding subtask module is assigned to the CPU computing intensive microservice pool; If the intensive computing type determination result is the second determination result, the corresponding subtask module will be assigned to the GPU computing-intensive microservice pool.

5. The voiceprint recognition method based on distributed heterogeneous computing according to claim 1, characterized in that, The formula for the dense computation judgment algorithm is as follows: Wherein, K represents the intensive computing type information, a represents the CPU utilization rate, x represents the CPU weight, b represents the memory utilization rate, y represents the memory weight, c represents the interface load, z represents the interface weight, d represents the GPU utilization rate, f represents the GPU weight, e represents the video memory utilization rate, and g represents the video memory weight.

6. The voiceprint recognition method based on distributed heterogeneous computing according to claim 1, characterized in that, The preset voiceprint recognition system is task-splitting to obtain multiple sub-task modules, including: Extract the corresponding sub-task module from the voiceprint recognition system according to the preset voiceprint recognition task; The sub-task modules include: format conversion module, channel separation module, noise reduction module, Mel frequency cepstral coefficient extraction module, frequency domain feature extraction module, cepstral mean normalization module, speech endpoint detection module, and voiceprint feature extraction module.

7. A voiceprint recognition device based on distributed heterogeneous computing, characterized in that, The device includes: The first processing module is used to acquire the speech signal to be recognized; The second processing module is used to split the preset voiceprint recognition system into multiple sub-task modules. The third processing module is used to evaluate each of the sub-task modules based on a preset intensive computing evaluation model to obtain the intensive computing type information corresponding to the sub-task module. The fourth processing module is used to divide the multiple sub-task modules into a preset CPU computing-intensive microservice pool and a preset GPU computing-intensive microservice pool based on the intensive computing type information corresponding to each sub-task module. The fifth processing module is used to perform a first calculation on the speech signal to be recognized based on the sub-task module of the CPU-intensive microservice pool to obtain first processing information; The sixth processing module is used to perform a second computational processing on the first processing information based on the sub-task module of the GPU computing-intensive microservice pool to obtain the voiceprint recognition result; The evaluation process for each sub-task module based on a preset intensive computing evaluation model, to obtain intensive computing type information corresponding to the sub-task module, includes: Each sub-task module undergoes unified interface processing and microservice transformation to obtain multiple microservice modules with unified interfaces. Based on the intensive computing evaluation model, a third computational process is performed on each microservice module to obtain judgment data corresponding to each microservice module; wherein, the judgment data includes CPU utilization, CPU weight, memory utilization, memory weight, interface load, interface weight, GPU utilization, GPU weight, video memory utilization, and video memory weight. Based on a preset intensive computing judgment algorithm, a fourth calculation process is performed on the CPU utilization rate, CPU weight, memory utilization rate, memory weight, interface load, interface weight, GPU utilization rate, GPU weight, video memory utilization rate, and video memory weight to obtain the intensive computing type information corresponding to the subtask module.

8. An electronic device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, when the processor executes the computer program, it implements the voiceprint recognition method based on distributed heterogeneous computing as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing computer-executable instructions for performing the voiceprint recognition method based on distributed heterogeneous computing as described in any one of claims 1 to 6.

Images