Data processing system based on voice recognition intelligent electronic scale

Abstract

The invention relates to a data processing system based on a voice recognition intelligent electronic scale, and belongs to the technical field of data processing. The method comprises a signal preprocessing module, a voice recognition module, a data updating and storage module, an inventory early warning module and a sales prediction module. The signal preprocessing module obtains an original audio signal and obtains a frame frequency characteristic through a preprocessing model according to the original audio signal. The speech recognition module obtains recognition content types through a speech recognition model according to the frame frequency characteristics; the data updating storage module obtains an updated commodity storage library and sales volume change data through an updating storage model according to the identification content type; the inventory early warning module obtains an inventory processing result through inventory judgment according to the updated commodity storage library; and the sales volume prediction module is used for obtaining a sales volume prediction result through a sales volume prediction model according to the sales volume change data. The data processing of the intelligent electronic scale based on voice recognition is realized.

Description

Technical Field

[0001] The invention belongs to the technical field of data processing, and in particular relates to a data processing system based on a voice recognition intelligent electronic scale. Background Art

[0002] At present, the sales and management of supermarkets mainly rely on the manual services of sales staff and the complicated and repetitive manual statistics of managers, resulting in low sales efficiency and prone to errors. The voice recognition smart electronic scale is an electronic weighing device that integrates voice recognition technology and intelligent functions. It can realize multi-functional operation through voice commands and can effectively solve the above problems. However, the large amount of data stored in the voice recognition smart electronic scale is often not fully utilized, and inventory management still mainly relies on short-term feedback from the market, making it difficult to achieve accurate sales forecasts.

[0003] Traditional sales forecasting methods based on statistics have been unable to adapt to the complex changes in today's sales market, and their forecasting accuracy and adaptability are low; although the single forecasting model based on neural networks can capture nonlinear relationships, it is prone to overfitting problems, resulting in low forecasting accuracy; sales forecasting methods based on combined model forecasting usually rely on historical sales data for data processing and analysis, but lack the ability to update sales data in real time and respond dynamically, making it difficult to meet the rapidly changing market needs. Therefore, the above methods still have shortcomings in real-time data collection, processing efficiency, and agile response to market trends. Summary of the invention

[0004] In order to solve the above problems existing in the prior art, the present invention provides a data processing system based on speech recognition intelligent electronic scale. The purpose of the present invention can be achieved through the following technical solutions: A data processing system based on voice recognition intelligent electronic scale, comprising a signal preprocessing module, a voice recognition module, a data update storage module, an inventory warning module, and a sales forecast module; The signal preprocessing module is used to obtain the original audio signal, and obtain the frame frequency feature according to the original audio signal through the preprocessing model; The speech recognition module is used to obtain the recognized content type through a speech recognition model according to the frame frequency feature; The data update storage module is used to obtain updated commodity storage and sales volume change data by updating the storage model according to the identified content type; The inventory warning module is used to obtain an inventory processing result through inventory judgment according to the updated commodity storage repository; The sales volume forecasting module is used to obtain sales volume forecasting results through a sales volume forecasting model according to the sales volume change data.

[0005] Preferably, obtaining the frame frequency feature through a preprocessing model according to the original audio signal includes: According to the original audio signal, noise reduction is performed by wavelet transform to obtain a noise-reduced audio signal; Pre-emphasize the noise-reduced audio signal through a first-order high-pass filter to obtain an emphasized high-frequency noise-reduced audio signal; According to the emphasized high-frequency noise reduction audio signal, a emphasized high-frequency stable noise reduction audio frame is obtained by frame division and windowing; Extracting frequency features through fast Fourier transform according to the aggravated high-frequency stable noise reduction frequency frame to obtain frame spectrum features; The frame frequency feature is obtained by converting the frame spectrum feature through a Mel filter.

[0006] Preferably, the identifying content type obtained by using a speech recognition model according to the frame frequency feature includes: Downsampling the frame frequency feature by convolution downsampling to obtain a downsampled frequency feature; Obtaining local features through a convolution enhancement model according to the down-sampling frequency features; Obtaining global features through a Fastformer model according to the downsampling frequency features; Merging the local features and the global features through depth-wise separable convolution to obtain fused features; Extracting residual features through a feedforward neural network based on the fusion features; The residual feature is added to the original audio signal to obtain the identified content type.

[0007] Preferably, obtaining the local features through a convolution enhancement model according to the downsampling frequency features comprises: Obtaining a group frequency feature by grouping according to the downsampling frequency feature; Obtaining grouped convolution features through multi-scale convolution according to the grouped frequency features; According to the grouped convolutional features, weight allocation is performed through the SENet model to obtain allocation weights; The local feature is obtained by adding the assigned weight to the down-sampled frequency feature.

[0008] Preferably, obtaining the global feature through the Fastformer model according to the downsampling frequency feature includes: Obtaining a query matrix, a key matrix, and a value matrix through linear calculation according to the downsampling frequency characteristics; Obtain a global token by token calculation according to the key matrix; Obtaining a global interaction output through global interaction according to the query matrix and the global token; Obtaining an element interaction output through element-level interaction according to the global interaction output and the value matrix; The global feature is obtained by restoring the original dimension through linear transformation according to the element interaction output.

[0009] Preferably, the updating of the commodity storage repository and sales volume change data by updating the storage model according to the identified content type includes: Acquire a commodity storage repository, collect weight data through a weight sensor, and update the commodity storage repository through the weight data and the identified content type to obtain the updated commodity storage repository; The commodity storage repository data and the updated commodity storage repository data are acquired, and the sales volume change data is obtained by performing a subtraction operation based on the commodity storage repository data and the updated commodity storage repository data.

[0010] Preferably, obtaining the inventory processing result through inventory judgment according to the updated commodity storage library includes: Preset an inventory threshold, obtain the updated commodity storage inventory, and determine whether the updated commodity storage inventory is greater than the inventory threshold. If yes, do not perform any operation; if no, send a replenishment signal to the smart terminal; Obtain the product shelf life from the updated product storage repository, obtain the current date, and determine whether the product shelf life exceeds the current date. If yes, send information about the existence of unsalable products to the smart terminal; if not, do not perform any operation.

[0011] Preferably, obtaining the sales forecast result through a sales forecast model according to the sales change data includes: Acquire weather data and holiday data, and obtain fuzzy weather data and fuzzy holiday data through fuzzification processing according to the weather data and the holiday data; Preset an empty data set, store the sales change data, the fuzzy weather data, and the fuzzy holiday data into the empty data set to obtain a sales data set; Obtaining a normalized sales data set by normalizing the sales data set; Obtaining normalized sales data from the normalized sales data set, and obtaining hyperparameters through the LSTM-GRU model according to the normalized sales data; According to the hyperparameters, an optimized hyperparameter is obtained by using a PSO optimization algorithm; Optimizing the LSTM-GRU model by optimizing the hyperparameters to obtain an optimized LSTM-GRU model; Predicting the normalized sales data using the optimized LSTM-GRU model to obtain a preliminary sales forecast result; The sales forecast result is obtained through the Dropout model based on the preliminary sales forecast result.

[0012] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, the data processing system based on speech recognition intelligent electronic scale is realized.

[0013] A storage medium containing computer executable instructions, wherein the computer executable instructions are used to execute the data processing system of the above-mentioned voice recognition intelligent electronic scale when executed by a computer processor.

[0014] The beneficial effects of the present invention are: (1) Wavelet transform is used to reduce noise, first-order high-pass filter is used to enhance high-frequency signals, frame division and windowing are used to improve signal stability, fast Fourier transform is used to extract signal spectrum features, and Mel filter conversion is used to emphasize important frequency ranges, thereby improving signal quality, reducing noise interference, and improving the speech recognition ability of the smart electronic scale.

[0015] (2) By combining the E-Branchformer model, the multi-scale deep convolutional neural network and the SENet model, the ability to capture and process detailed features is enhanced while performing speech recognition, thereby improving the accuracy and robustness of speech recognition.

[0016] (3) By updating inventory information and obtaining sales change data, accurate sales forecasts can be made based on sales change data, weather data, and holiday data. This can help predict future sales trends, conduct real-time inventory management, improve processing efficiency, and respond quickly to market trends, helping companies achieve better business planning and sales strategies and reduce resource waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0018] Figure 1 The present invention is a flow chart of a data processing system based on speech recognition intelligent electronic scale. DETAILED DESCRIPTION

[0019] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0020] See also Figure 1, a data processing system based on speech recognition intelligent electronic scale, including a signal preprocessing module, a speech recognition module, a data update storage module, an inventory warning module, and a sales forecast module; The signal preprocessing module is used to obtain the original audio signal, and obtain the frame frequency feature according to the original audio signal through the preprocessing model; The speech recognition module is used to obtain the recognized content type through a speech recognition model according to the frame frequency feature; The data update storage module is used to obtain updated commodity storage and sales volume change data by updating the storage model according to the identified content type; The inventory warning module is used to obtain an inventory processing result through inventory judgment according to the updated commodity storage repository; The sales volume forecasting module is used to obtain sales volume forecasting results through a sales volume forecasting model according to the sales volume change data.

[0021] In this embodiment, the voice recognition intelligent electronic scale includes a weight sensor, a voice recognition module, an STM32 processing unit, a communication module, and a display screen; the voice recognition intelligent electronic scale can realize multi-functional operation through voice commands, and the voice commands include weighing mode switching, data entry, and result broadcasting; the voice recognition intelligent electronic scale can combine the Internet of Things and data analysis technology to manage and apply weighing data.

[0022] Specifically, the preprocessing model includes: S101: reducing noise by wavelet transform according to the original audio signal to obtain a reduced-noise audio signal; S102: pre-emphasize the noise-reduced audio signal through a first-order high-pass filter to obtain an emphasized high-frequency noise-reduced audio signal; S103: Obtaining a weighted high-frequency stable noise reduction audio frame by performing frame division and windowing according to the weighted high-frequency noise reduction audio signal; S104: extracting frequency features through fast Fourier transform according to the high-frequency stabilized noise reduction frequency frame to obtain frame spectrum features; S105: Obtain the frame frequency feature through Mel filter conversion according to the frame spectrum feature.

[0023] Specifically, the pre-emphasis means improving the high-frequency signal capability and enhancing the characteristics of the high-frequency part.

[0024] Specifically, the speech recognition model includes: S201: downsampling the frame frequency feature by convolution downsampling to obtain a downsampling frequency feature; S202: Obtaining local features through a convolution enhancement model according to the downsampling frequency features; S203: Obtaining global features through a Fastformer model according to the downsampling frequency features; S204: merging the local features and the global features through depthwise separable convolution to obtain fused features; Specifically, the merging expression of the depth-separable convolution blocks is: , in, Y C is a connection feature, wherein the connection feature represents a feature obtained by connecting the global feature and the local feature, Y G is the global feature, Y L is the local feature, Y D represents performing the depth-wise separable convolution on the connection feature, Y Merge represents the fusion feature, Concat Indicates a connection operation. DwConv represents the depthwise separable convolution, W Represents feature projection; S205: extracting residual features through a feedforward neural network according to the fusion features; S206: Add the residual feature to the original audio signal to obtain the identified content type.

[0025] Specifically, the convolution enhancement model includes: S202-1: Obtaining a group frequency feature by grouping according to the downsampling frequency feature; S202-2: Obtaining group convolution features through multi-scale convolution according to the group frequency features; S202-3: performing weight allocation through the SENet model according to the grouped convolutional features to obtain allocation weights; S202-4: Obtain the local feature by adding the allocated weight to the down-sampled frequency feature; The local feature is expressed as: , in, Out is the local feature, X r is the downsampling frequency characteristic, SE_Connect Indicates that the SENet model performs weight distribution, X merge is the grouped convolution feature, SE_Connect(X merge )represents the allocation weight.

[0026] Specifically, the Fastformer model includes: S203-1: Obtain a query matrix, a key matrix, and a value matrix through linear calculation according to the downsampling frequency characteristics; The linear calculation expression is: , in, Q is the query matrix, X is the downsampling frequency characteristic, W Q , W K , W V is the weight matrix, K is the bond matrix, V is the value matrix; S203-2: Obtain a global token through token calculation according to the key matrix; The token calculation expression is: , in, g is the global token, N represents the number of key matrices, K i represents the i-th bond matrix; S203-3: Obtaining a global interaction output through global interaction according to the query matrix and the global token; The global interaction expression is: , in, U i is the global interaction output, σ is the activation function, represents element-wise multiplication, Q i represents the i-th query matrix, g is the global token; S203-4: Obtaining an element interaction output through element-level interaction according to the global interaction output and the value matrix; The element-level interaction expression is:

[0027] in, O i Indicates the interactive output of the elements, U i is the global interaction output, Vi is the i-th value matrix; S203-5: restoring the original dimension through linear transformation according to the element interaction output to obtain the global feature; The linear variation expression is: , in, Y is the global feature, O Interactive output for said element, W O is the weight matrix.

[0028] Specifically, the updating storage model includes: S301: Acquire a commodity storage repository, collect weight data through a weight sensor, and update the commodity storage repository through the weight data and the identified content type to obtain the updated commodity storage repository; S302: Acquire commodity storage repository data and update commodity storage repository data, and obtain the sales volume change data through a subtraction operation based on the commodity storage repository data and the updated commodity storage repository data.

[0029] Specifically, the inventory determination includes: S401: Preset an inventory threshold, obtain the inventory of the updated commodity storage repository, and determine whether the inventory of the updated commodity storage repository is greater than the inventory threshold. If yes, do not perform any operation; if no, send a replenishment signal to the smart terminal; S402: Obtain the shelf life of the product from the updated product storage repository, obtain the current date, and determine whether the shelf life of the product exceeds the current date. If yes, send information about the existence of unsaleable products to the smart terminal; if not, do not perform any operation.

[0030] Specifically, the sales forecasting model includes: S501: Acquire weather data and holiday data, and obtain fuzzy weather data and fuzzy holiday data through fuzzification processing according to the weather data and the holiday data; S502: Preset an empty data set, and store the sales change data, the fuzzy weather data, and the fuzzy holiday data into the empty data set to obtain a sales data set; S503: Obtaining a normalized sales data set by normalizing the sales data set; S504: Obtaining normalized sales data from the normalized sales data set, and obtaining hyperparameters through the LSTM-GRU model according to the normalized sales data; S505: Obtaining optimized hyperparameters through a PSO optimization algorithm according to the hyperparameters; S506: Optimizing the LSTM-GRU model by optimizing the hyperparameters to obtain an optimized LSTM-GRU model; S507: Predicting the normalized sales data using the optimized LSTM-GRU model to obtain a preliminary sales forecast result; S508: Obtain the sales forecast result through the Dropout model according to the preliminary sales forecast result.

[0031] The computer storage medium of the embodiment of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device.

[0032] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0033] The program code included on the computer readable medium can be transmitted with any appropriate medium, including but not limited to wireless, electric wire, optical cable, RF, etc., or any suitable combination of the above. The computer program code for performing the operation of the present invention can be written in one or more programming languages ​​or their combinations, and the programming language includes object-oriented programming languages-such as Java, Smalltalk, C++, and also includes conventional procedural programming languages-such as "C" language or similar programming languages. The program code can be executed completely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on the remote computer, or completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0034] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A data processing system based on speech recognition intelligent electronic scale, characterized in that: It includes signal preprocessing module, speech recognition module, data update storage module, inventory warning module and sales forecast module; The signal preprocessing module is used to obtain the original audio signal, and obtain the frame frequency feature according to the original audio signal through the preprocessing model; The speech recognition module is used to obtain the recognized content type through a speech recognition model according to the frame frequency feature; The data update storage module is used to obtain updated commodity storage and sales volume change data by updating the storage model according to the identified content type; The inventory warning module is used to obtain an inventory processing result through inventory judgment according to the updated commodity storage repository; The sales volume forecasting module is used to obtain sales volume forecasting results through a sales volume forecasting model according to the sales volume change data.

2. The data processing system based on speech recognition intelligent electronic scale according to claim 1 is characterized in that: The obtaining of the frame frequency feature by the preprocessing model according to the original audio signal comprises: According to the original audio signal, noise reduction is performed by wavelet transform to obtain a noise-reduced audio signal; Pre-emphasize the noise-reduced audio signal through a first-order high-pass filter to obtain an emphasized high-frequency noise-reduced audio signal; According to the emphasized high-frequency noise reduction audio signal, a emphasized high-frequency stable noise reduction audio frame is obtained by frame division and windowing; Extracting frequency features through fast Fourier transform according to the aggravated high-frequency stable noise reduction frequency frame to obtain frame spectrum features; The frame frequency feature is obtained by converting the frame spectrum feature through a Mel filter.

3. The data processing system based on speech recognition intelligent electronic scale according to claim 1 is characterized in that: The type of content identified by the speech recognition model according to the frame frequency feature includes: Downsampling the frame frequency feature by convolution downsampling to obtain a downsampled frequency feature; Obtaining local features through a convolution enhancement model according to the down-sampling frequency features; Obtaining global features through a Fastformer model according to the downsampling frequency features; Merging the local features and the global features through depth-wise separable convolution to obtain fused features; Extracting residual features through a feedforward neural network based on the fusion features; The residual feature is added to the original audio signal to obtain the identified content type.

4. The data processing system based on speech recognition intelligent electronic scale according to claim 3 is characterized in that: The obtaining of local features through a convolution enhancement model according to the down-sampling frequency features comprises: Obtaining a group frequency feature by grouping according to the downsampling frequency feature; Obtaining grouped convolution features through multi-scale convolution according to the grouped frequency features; According to the grouped convolutional features, weight allocation is performed through the SENet model to obtain allocation weights; The local feature is obtained by adding the assigned weight to the down-sampled frequency feature.

5. The data processing system based on speech recognition intelligent electronic scale according to claim 3 is characterized in that: The obtaining of the global feature through the Fastformer model according to the downsampling frequency feature includes: Obtaining a query matrix, a key matrix, and a value matrix through linear calculation according to the downsampling frequency characteristics; Obtain a global token by token calculation according to the key matrix; Obtaining a global interaction output through global interaction according to the query matrix and the global token; Obtaining an element interaction output through element-level interaction according to the global interaction output and the value matrix; The global feature is obtained by restoring the original dimension through linear transformation according to the element interaction output.

6. The data processing system based on speech recognition intelligent electronic scale according to claim 1 is characterized in that: The updating of the commodity storage repository and sales volume change data by updating the storage model according to the identified content type includes: Acquire a commodity storage repository, collect weight data through a weight sensor, and update the commodity storage repository through the weight data and the identified content type to obtain the updated commodity storage repository; The commodity storage repository data and the updated commodity storage repository data are acquired, and the sales volume change data is obtained by performing a subtraction operation based on the commodity storage repository data and the updated commodity storage repository data.

7. The data processing system based on speech recognition intelligent electronic scale according to claim 1 is characterized in that: The obtaining of the inventory processing result by inventory judgment according to the updating of the commodity storage library comprises: Preset an inventory threshold, obtain the updated commodity storage inventory, and determine whether the updated commodity storage inventory is greater than the inventory threshold. If yes, do not perform any operation; if no, send a replenishment signal to the smart terminal; Obtain the product shelf life from the updated product storage repository, obtain the current date, and determine whether the product shelf life exceeds the current date. If yes, send information about the existence of unsalable products to the smart terminal; if not, do not perform any operation.

8. The data processing system based on speech recognition intelligent electronic scale according to claim 1 is characterized in that: The obtaining of sales volume forecast results by using a sales volume forecast model according to the sales volume change data includes: Acquire weather data and holiday data, and obtain fuzzy weather data and fuzzy holiday data through fuzzification processing according to the weather data and the holiday data; Preset an empty data set, store the sales change data, the fuzzy weather data, and the fuzzy holiday data into the empty data set to obtain a sales data set; Obtaining a normalized sales data set by normalizing the sales data set; Obtaining normalized sales data from the normalized sales data set, and obtaining hyperparameters through the LSTM-GRU model according to the normalized sales data; According to the hyperparameters, an optimized hyperparameter is obtained by using a PSO optimization algorithm; Optimizing the LSTM-GRU model by optimizing the hyperparameters to obtain an optimized LSTM-GRU model; Predicting the normalized sales data using the optimized LSTM-GRU model to obtain a preliminary sales forecast result; The sales forecast result is obtained through the Dropout model based on the preliminary sales forecast result.

9. 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 program, the data processing system based on the speech recognition intelligent electronic scale as described in any one of claims 1 to 8 is implemented.

10. A storage medium containing computer executable instructions, characterized in that: The computer executable instructions are used to execute the data processing system based on voice recognition intelligent electronic scale as described in any one of claims 1 to 8 when executed by a computer processor.

Images