A beverage intelligent blending method and system based on emotion recognition

Abstract

The application relates to the technical field of intelligent beverage blending and man-machine interaction, and provides a beverage intelligent blending method based on emotion recognition, which comprises the following steps: obtaining voice input of a user, converting the voice input into text through voice recognition; performing emotion analysis on the text by using a language model, extracting at least two emotion dimensions and determining an emotion state of the user; according to a preset emotion-flavor mapping mechanism, mapping the emotion state into a corresponding flavor combination, and generating a formula proportion of the flavor combination; controlling a beverage blending device to execute blending according to the formula proportion, and outputting a corresponding beverage. Through the introduction of a trained language model for emotion recognition of user voice, and in combination with a preset emotion-flavor mapping mechanism, a leap from traditional taste customization to emotion-driven individualization of beverages is realized, the emotional added value of the beverage is improved, thousands of different flavor combinations are generated through the proportion change of flavors, and thus the individualized needs of thousands of people with thousands of tastes and thousands of flavors with thousands of emotions are met.

Description

Technical Field

[0001] This invention relates to the field of intelligent beverage preparation and human-computer interaction technology, and in particular to an intelligent beverage preparation method and system based on emotion recognition. Background Technology

[0002] Current intelligent beverage blending systems primarily rely on preset fixed recipes or user-selected flavor combinations to achieve a certain degree of personalized beverage customization. However, these customization methods mostly remain at the functional level of ingredient adjustments, failing to reflect the user's immediate emotional and psychological needs.

[0003] With the increasing maturity of applications such as speech recognition, natural language processing (NLP), and large language models (LLM) in emotion recognition and human-computer interaction, research shows that information such as speech tone and text content can effectively reflect a user's emotional state. If emotion recognition could be combined with beverage blending systems to generate corresponding flavor ratios based on the user's real-time mood, theoretically, it could significantly enhance the personalization and emotional value of beverage experiences. However, there is currently no complete solution in the technology that combines speech transcription, emotion analysis, and dynamic generation of beverage recipes. Existing beverage machines and unmanned coffee kiosks lack emotion-driven intelligent dispensing capabilities, and cannot achieve true personalization of "a thousand people, a thousand flavors; a thousand emotions, a thousand tastes". Summary of the Invention

[0004] To address the shortcomings of existing beverage blending systems that rely solely on fixed recipes or manual user selection and lack personalized responses based on emotional states, this invention aims to provide an intelligent beverage blending method and system based on emotion recognition. Through voice recognition and emotion analysis, the system maps the user's real-time emotional state to flavor formula proportions, thereby achieving emotional and personalized beverage customization, enhancing the user's interactive experience and emotional value. This invention achieves its objective through the following technical solutions: This invention provides a smart beverage blending method based on emotion recognition, comprising: Acquire user voice input and transcribe it into text using speech recognition; The text is analyzed for sentiment using a trained language model to extract at least two sentiment dimensions and determine the user's emotional state. Based on the preset emotion-flavor mapping mechanism, the emotional state is mapped to the corresponding flavor combination, and the recipe ratio of the flavor combination is generated. The beverage mixing equipment is controlled to mix beverages according to the recipe proportions and output the corresponding beverages.

[0005] Furthermore, using a trained language model, sentiment analysis is performed on the text to extract at least two sentiment dimensions and determine the user's emotional state, including... The text is cleaned, corrected, segmented, and denoised to obtain analytical text. The analyzed text is input into the trained language model to obtain the sentiment feature vector corresponding to the analyzed text. Based on emotion feature vector regression or by calculating the values ​​of at least two emotion dimensions; Based on the position of the emotion dimension in the preset emotion space, emotion labels and / or quadrant categories are generated to determine the emotion state.

[0006] Furthermore, the emotion dimension includes pleasure and arousal; pleasure is used to characterize the positive or negative tendency of a user's emotions, while arousal is used to characterize the level of activity of a user's emotions.

[0007] Furthermore, the preset emotion space is a two-dimensional emotion space constructed using two indicators: pleasure and arousal. The two-dimensional emotion space is divided into four quadrants. Based on the emotion feature vector of the analyzed text, the position in the two-dimensional emotion space is calculated, the emotion state is determined to belong to one of the above quadrants, and mapped to the corresponding flavor category.

[0008] Furthermore, the four quadrants are high pleasure and high arousal, high pleasure and low arousal, low pleasure and high arousal, and low pleasure and low arousal; emotional state corresponds to one area in the four quadrants.

[0009] Furthermore, the emotional state is determined to belong to one of the above quadrants and mapped to the corresponding flavor category, including: Read the values ​​of pleasure and arousal from the emotion feature vector; Based on a comparison of preset thresholds for pleasure and arousal, a quadrant index is obtained. The corresponding flavor category is then retrieved from a stored emotion-flavor mapping table using this index, and the flavor category is output as an input parameter for subsequent recipe proportion generation. Flavor categories include fruity flavors, floral flavors, nutty and vanilla flavors, and blended flavors.

[0010] Furthermore, the formulation proportions for generating flavor combinations include, The target flavor category is determined based on the emotional state, and at least one specific flavor is selected as the main flavor and at least one specific flavor as the auxiliary flavor in the target flavor category. Select at least one blending flavor to balance the overall flavor; The recipe uses a pre-defined ratio of main flavor, secondary flavor, and blending flavor, with the main flavor having a higher proportion than the secondary flavor, and the secondary flavor having a higher proportion than the blending flavor.

[0011] Furthermore, generating the formula ratio of flavor combination also includes introducing random perturbations within a preset range based on the formula ratio, so as to increase the diversity of flavor combination while maintaining the overall proportional relationship, and output the formula ratio of flavor combination as input parameters for beverage blending equipment to perform blending.

[0012] Furthermore, the beverage mixing equipment is controlled to mix beverages according to the formula ratio and output the corresponding beverages, including... The formula ratio is converted into execution instructions for beverage mixing equipment. The execution instructions include the dispensing time or dispensing volume parameters for each flavor syrup pump. The basic liquid supply unit controls the output of beverage base liquid, which is mixed with flavor syrup in proportion. The beverage base liquid includes at least one of coffee liquid, milk or tea liquid. The mixing unit or flow channel mixing unit is synchronously controlled to ensure that the base liquid and various flavored syrups are fully and evenly mixed. The mixing operation is terminated when the beverage container reaches the preset capacity threshold, and the beverage is output.

[0013] Based on the same inventive concept, this invention provides an intelligent beverage blending system, employing the intelligent beverage blending method described above, including: The acquisition module is used to acquire the user's voice input and transcribe the voice input into text through speech recognition; The sentiment analysis module is used to perform sentiment analysis on text using a trained language model, extract at least two sentiment dimensions, and determine the user's emotional state. This includes cleaning, correcting errors, segmenting, and denoising the text to obtain the analysis text; inputting the analysis text into the trained language model to obtain a sentiment feature vector corresponding to the analysis text; regressing or calculating the values ​​of at least two sentiment dimensions based on the sentiment feature vector; and generating sentiment labels and / or quadrant categories based on the position of the sentiment dimensions in a preset sentiment space to determine the emotional state. The recipe generation module maps emotional states to corresponding flavor combinations based on a preset emotion-flavor mapping mechanism and generates the recipe proportions for these flavor combinations. This includes reading the values ​​of pleasure and arousal from the emotional feature vector; comparing these values ​​against preset thresholds to obtain a quadrant index; retrieving the corresponding flavor category from a stored emotion-flavor mapping table based on the quadrant index; and outputting the flavor category as input parameters for subsequent recipe proportion generation. Flavor categories include fruity, floral, nutty and vanilla, and blended flavors. The module determines the target flavor category based on the emotional state and selects at least one specific flavor as the main flavor and at least one specific flavor as an auxiliary flavor from each target flavor category. It also selects at least one blended flavor for overall flavor balance and calculates the recipe proportions of the main flavor, auxiliary flavor, and blended flavor according to a preset ratio, where the proportion of the main flavor is greater than that of the auxiliary flavor, and the proportion of the auxiliary flavor is greater than that of the blended flavor. The beverage blending control module is used to control the beverage blending equipment to blend beverages according to the formula ratio and output the corresponding beverages. This includes converting the formula ratio into execution instructions for the beverage blending equipment, which include parameters for the dispensing time or volume of each flavor syrup pump; controlling the basic liquid supply unit to provide the dispensing volume of the beverage base liquid, which is mixed with the flavor syrups in proportion. The beverage base liquid includes at least one of coffee liquid, milk, or tea liquid; synchronously controlling the stirring unit or the flow channel mixing unit to ensure that the base liquid and each flavor syrup are fully and evenly mixed; terminating the blending operation when the beverage container reaches a preset capacity threshold, and outputting the beverage.

[0014] Compared with the prior art, the present invention has at least one of the following beneficial effects: This invention introduces a trained language model to recognize user voice emotions and combines it with a pre-set emotion-flavor mapping mechanism to achieve a leap from traditional flavor customization to emotion-driven personalization of beverages, enhancing the emotional value of drinks. It generates thousands of different flavor combinations through variations in flavor proportions, thus satisfying the personalized needs of a thousand people with a thousand tastes and a thousand emotions. Simultaneously, the system uses voice interaction, effectively lowering the user's operational threshold and improving the naturalness and immersion of human-computer interaction. Furthermore, the system possesses self-learning and dynamic optimization capabilities, continuously adjusting flavor ratios based on user feedback to achieve a continuously optimized customized experience and enhance long-term user stickiness. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the steps of the intelligent beverage blending method based on emotion recognition according to the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0017] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this specification means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0018] First Embodiment In practical applications within the beverage industry, existing personalized blending systems primarily rely on users actively selecting recipes, such as choosing flavor combinations through menus or interfaces. While this approach functionally satisfies the need for "flavor diversity," it remains focused on customizing the physical taste and lacks consideration for the user's psychological and emotional state. When users are in different emotional states such as tension, depression, or excitement, relying solely on preset flavor selections often fails to convey effective emotional value or create a deep user experience.

[0019] Based on this pain point, the inventors conducted in-depth research and realized that emotions can be accurately identified by artificial intelligence technology and can also be mapped to the flavor characteristics of beverages. Therefore, they proposed an intelligent beverage blending solution based on emotion recognition: user input is acquired through voice interaction, a trained language model is used to analyze the emotional dimension, mapping the user's psychological state to a corresponding flavor combination, and then the beverage blending equipment automatically generates the corresponding recipe proportions. This technical solution not only breaks through the limitations of the traditional beverage system's "functionality" but also endows beverages with the ability to "respond to emotions," allowing users to experience a personalized experience tailored to their current mood with each drink. The specific implementation method is as follows: like Figure 1 As shown, this invention provides a smart beverage blending method based on emotion recognition, including: Step S1: The user interacts with the beverage preparation system via voice, obtains the user's voice input, and transcribes the voice input into text through voice recognition; then, ASR (Automatic Speech Recognition) is used to transcribe the voice information into text. Step S2: Use the trained language model to perform sentiment analysis on the text, extract at least two sentiment dimensions, and determine the user's emotional state; Step S3: Based on the preset emotion-flavor mapping mechanism, map the emotional state to the corresponding flavor combination and generate the recipe ratio of the flavor combination; Step S4: Control the beverage mixing equipment to mix the beverages according to the recipe ratio and output the corresponding beverages.

[0020] Furthermore, using a trained language model, sentiment analysis is performed on the text to extract at least two sentiment dimensions and determine the user's emotional state, including... The text is cleaned, corrected, segmented, and denoised to obtain analytical text. The analyzed text is input into the trained language model to obtain the sentiment feature vector corresponding to the analyzed text. Based on emotion feature vector regression or by calculating the values ​​of at least two emotion dimensions; Based on the position of the emotion dimension in the preset emotion space, emotion labels and / or quadrant categories are generated to determine the emotion state.

[0021] Specifically, the original text obtained through ASR transcription is cleaned, corrected, segmented, and denoised to eliminate colloquialisms, typos, stop words, and repeated words, resulting in a structured analytical text. This analytical text is then input into a trained language model, such as a large-scale semantic model based on the Transformer architecture. The model outputs a corresponding emotion feature vector through word vector encoding, context modeling, and attention mechanisms. This vector contains high-dimensional numerical features related to emotion.

[0022] Based on the aforementioned emotion feature vector, regression calculation or classification reasoning methods are used to obtain numerical values ​​for at least two emotion dimensions; these emotion dimensions include, but are not limited to, Valence: Indicates the positive or negative tendency of emotions, and the numerical range is usually [−1,1] or [0,1]. Arousal: Indicates the level of emotional activity or excitement, with a value typically ranging from [0,1]. Based on the combined position of pleasure and arousal in the two-dimensional emotion space, corresponding emotion labels (such as "happy", "nervous", "fatigued") and / or quadrant categories (such as high pleasure × high arousal, low pleasure × low arousal, etc.) are generated to determine the user's emotional state.

[0023] Furthermore, the emotion dimension includes pleasure and arousal; pleasure is used to characterize the positive or negative tendency of a user's emotions, while arousal is used to characterize the level of activity of a user's emotions.

[0024] Furthermore, the preset emotion space is a two-dimensional emotion space constructed using two indicators: pleasure and arousal. The two-dimensional emotion space is divided into four quadrants. Based on the emotion feature vector of the analyzed text, the position in the two-dimensional emotion space is calculated, the emotion state is determined to belong to one of the above quadrants, and mapped to the corresponding flavor category.

[0025] Furthermore, the four quadrants are high pleasure and high arousal, high pleasure and low arousal, low pleasure and high arousal, and low pleasure and low arousal; emotional state corresponds to one area in the four quadrants.

[0026] Furthermore, the four quadrants correspond to the following emotion labels: Emotional states corresponding to high pleasure and high arousal include happiness, excitement, or anticipation; Emotional states corresponding to high pleasure and low arousal include gentleness, romance, or relaxation; Low pleasure and high arousal correspond to emotional states such as anxiety, tension, or unease. Emotional states associated with low pleasure and low arousal include fatigue, loss, or loneliness.

[0027] Furthermore, the emotional state is determined to belong to one of the above quadrants and mapped to the corresponding flavor category, including: Read the values ​​of pleasure and arousal from the emotion feature vector; Based on a comparison of preset thresholds for pleasure and arousal, a quadrant index is obtained. The corresponding flavor category is then retrieved from a stored emotion-flavor mapping table using this index, and the flavor category is output as an input parameter for subsequent recipe proportion generation. Flavor categories include fruity flavors, floral flavors, nutty and vanilla flavors, and blended flavors.

[0028] There is a mapping relationship between emotional state and flavor category. For example, floral flavors, such as magnolia, osmanthus, and camellia, are suitable for expressing delicate, soothing, or quiet emotions; fruity flavors, such as strawberry and honeydew melon, are suitable for expressing energetic, positive, or expectant emotions; nutty and vanilla flavors, such as hazelnut and vanilla, are suitable for expressing healing, warm, or reassuring emotions; and cane sugar, as a neutral blending flavor, can be flexibly added to different recipes to balance the overall taste.

[0029] Specifically, the system parses pleasure and arousal values ​​from the emotion feature vector. Pleasure represents the positive or negative tendency of a user's emotion, while arousal represents the level of activity of that emotion. The pleasure value is compared with a preset pleasure threshold to determine whether the emotion is high or low pleasure; simultaneously, the arousal value is compared with a preset arousal threshold to determine whether the emotion is high or low arousal. By combining these two determinations, a corresponding quadrant index is generated. For example, high pleasure and high arousal correspond to quadrant "11", high pleasure and low arousal to quadrant "10", low pleasure and high arousal to quadrant "01", and low pleasure and low arousal to quadrant "00". Based on this, the system searches a pre-stored emotion-flavor mapping table according to the quadrant index to obtain the target flavor category corresponding to the quadrant. For example, quadrant "11" corresponds to fruity flavors, quadrant "10" corresponds to floral flavors, quadrant "01" corresponds to nutty and vanilla flavors, and quadrant "00" corresponds to harmonious flavors. Finally, the flavor category is output and used as an input parameter for the subsequent recipe ratio generation module to guide the calculation of the proportions of the main flavor, auxiliary flavor, and blending flavor and the beverage preparation.

[0030] Regarding the allocation of the formula proportions, this embodiment provides a preferred strategy: a specific flavor from the target flavor category is used as the main flavor, accounting for approximately 50%; another flavor complementary to the main flavor is selected as an auxiliary flavor, accounting for approximately 30%; and a harmonizing flavor, such as sucrose or vanilla, is introduced, accounting for approximately 20%, to balance the overall taste. Through this method, flexible generation of flavor combinations can be achieved, satisfying the requirement that "different users in the same emotional state receive slightly different but overall consistent beverage experiences."

[0031] For example, in a specific application scenario, a user interacts with a beverage preparation system via voice input, such as saying, "I'm very nervous lately, I have an exam tomorrow." First, the Automatic Speech Recognition (ASR) module transcribes the speech information into text. Then, a trained language model is used to perform sentiment analysis on the text, obtaining the corresponding sentiment feature vector, and calculating the values ​​of pleasure and arousal. Based on the comparison, the user's emotional state is determined to be low pleasure and high arousal, with the sentiment label "nervous."

[0032] Based on a pre-defined emotion-flavor mapping table, the emotional state of low pleasure and high arousal is mapped to floral flavors. The system further selects camellia as the primary flavor (approximately 50%), osmanthus as a secondary flavor (approximately 30%), and introduces sucrose as a harmonizing flavor (approximately 20%), thus generating a complete flavor combination formula. To enhance the personalized experience, a random perturbation of ±0~10% can be added to the proportions of the primary and secondary flavors when generating the formula, creating slightly differentiated beverage flavors. When the formula gains high acceptance among users (e.g., the number of times it is shared via QR code and repeatedly made reaches a preset threshold), the formula is recorded and updated to the formula database for subsequent user access.

[0033] The formula ratio is sent to the beverage mixing equipment, which controls the syrup pump and base liquid supply to perform the mixing process, producing a beverage with camellia and osmanthus as the main ingredients and sucrose added to balance the taste. This aims to relieve tension and enhance the user's drinking experience. After a user finishes drinking the beverage, they can provide feedback through an interactive interface. This feedback includes taste satisfaction, flavor preference, and whether they would like to choose the recipe again. This feedback is associated with the user's current emotional state and the corresponding beverage recipe proportions, forming a feedback sample dataset. Based on this, the emotion-flavor mapping is continuously improved through statistical optimization and model optimization. In statistical optimization, when the proportion of positive feedback for a flavor recipe under a certain emotional state exceeds a preset threshold, the priority of that recipe in subsequent calls is increased; conversely, when the proportion of negative feedback for a recipe exceeds the threshold, the system reduces its weight or even removes the recipe. In optimization, the feedback sample data is input into the training module to iteratively update the emotion-flavor mapping mechanism, making the flavor weights more closely reflect the actual feelings of the user group. Through these methods, self-learning and dynamic upgrades are achieved based on user feedback, thereby gradually improving the personalization of beverage preparation and user stickiness in the long term.

[0034] Furthermore, the formulation proportions for generating flavor combinations include, The target flavor category is determined based on the emotional state, and at least one specific flavor is selected as the main flavor and at least one specific flavor as the auxiliary flavor in the target flavor category. Select at least one blending flavor to balance the overall flavor; The recipe uses a pre-defined ratio of main flavor, secondary flavor, and blending flavor, with the main flavor having a higher proportion than the secondary flavor, and the secondary flavor having a higher proportion than the blending flavor.

[0035] Furthermore, generating the formula ratio of flavor combination also includes introducing random perturbations within a preset range based on the formula ratio, so as to increase the diversity of flavor combination while maintaining the overall proportional relationship, and output the formula ratio of flavor combination as input parameters for beverage blending equipment to perform blending.

[0036] Furthermore, the beverage mixing equipment is controlled to mix beverages according to the formula ratio and output the corresponding beverages, including... The formula ratio is converted into execution instructions for beverage mixing equipment. The execution instructions include the dispensing time or dispensing volume parameters for each flavor syrup pump. The basic liquid supply unit controls the output of beverage base liquid, which is mixed with flavor syrup in proportion. The beverage base liquid includes at least one of coffee liquid, milk or tea liquid. The mixing unit or flow channel mixing unit is synchronously controlled to ensure that the base liquid and various flavored syrups are fully and evenly mixed. The mixing operation is terminated when the beverage container reaches the preset capacity threshold, and the beverage is output.

[0037] Second Embodiment Based on the same inventive concept, this invention provides an intelligent beverage blending system, employing the intelligent beverage blending method described above, including: The acquisition module is used to acquire the user's voice input and transcribe the voice input into text through speech recognition; The sentiment analysis module is used to perform sentiment analysis on text using a trained language model, extract at least two sentiment dimensions, and determine the user's emotional state. This includes cleaning, correcting errors, segmenting, and denoising the text to obtain the analysis text; inputting the analysis text into the trained language model to obtain a sentiment feature vector corresponding to the analysis text; regressing or calculating the values ​​of at least two sentiment dimensions based on the sentiment feature vector; and generating sentiment labels and / or quadrant categories based on the position of the sentiment dimensions in a preset sentiment space to determine the emotional state. The recipe generation module maps emotional states to corresponding flavor combinations based on a preset emotion-flavor mapping mechanism and generates the recipe proportions for these flavor combinations. This includes reading the values ​​of pleasure and arousal from the emotional feature vector; comparing these values ​​against preset thresholds to obtain a quadrant index; retrieving the corresponding flavor category from a stored emotion-flavor mapping table based on the quadrant index; and outputting the flavor category as input parameters for subsequent recipe proportion generation. Flavor categories include fruity, floral, nutty and vanilla, and blended flavors. The module determines the target flavor category based on the emotional state and selects at least one specific flavor as the main flavor and at least one specific flavor as an auxiliary flavor from each target flavor category. It also selects at least one blended flavor for overall flavor balance and calculates the recipe proportions of the main flavor, auxiliary flavor, and blended flavor according to a preset ratio, where the proportion of the main flavor is greater than that of the auxiliary flavor, and the proportion of the auxiliary flavor is greater than that of the blended flavor. The beverage blending control module is used to control the beverage blending equipment to blend beverages according to the formula ratio and output the corresponding beverages. This includes converting the formula ratio into execution instructions for the beverage blending equipment, which include parameters for the dispensing time or volume of each flavor syrup pump; controlling the basic liquid supply unit to provide the dispensing volume of the beverage base liquid, which is mixed with the flavor syrups in proportion. The beverage base liquid includes at least one of coffee liquid, milk, or tea liquid; synchronously controlling the stirring unit or the flow channel mixing unit to ensure that the base liquid and each flavor syrup are fully and evenly mixed; terminating the blending operation when the beverage container reaches a preset capacity threshold, and outputting the beverage.

[0038] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

[0039] It should be noted that the above embodiments can be freely combined as needed. The above description is only a preferred embodiment of the present invention. It should be pointed out that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for intelligent preparation of a beverage based on emotion recognition, characterized in that, The method comprises: obtaining a voice input of a user and transcribing the voice input into text through voice recognition; performing emotion analysis on the text by using a trained language model, extracting at least two emotion dimensions, and determining an emotion state of the user; mapping the emotion state to a corresponding flavor combination according to a preset emotion-flavor mapping mechanism, and generating a recipe proportion of the flavor combination; controlling a beverage preparation device to perform preparation according to the recipe proportion, and outputting a corresponding beverage.

2. The method of claim 1, wherein, The emotion analysis on the text by using a trained language model, the extraction of at least two emotion dimensions, and the determination of the emotion state of the user comprise: cleaning, correcting, segmenting, and denoising the text to obtain an analysis text; inputting the analysis text into a trained language model to obtain an emotion feature vector corresponding to the analysis text; regressing or calculating numerical values of at least two emotion dimensions based on the emotion feature vector; generating an emotion label and / or a quadrant category according to the position of the emotion dimensions in a preset emotion space, and determining the emotion state.

3. The method of claim 2, wherein, The emotion dimensions include a pleasantness and an arousal; wherein the pleasantness is used to represent the positive or negative tendency of the user's emotion, and the arousal is used to represent the energy level of the user's emotion.

4. The intelligent beverage formulation method of claim 3, wherein, The preset emotion space is a two-dimensional emotion space constructed by using the two indicators of the pleasantness and the arousal, and the two-dimensional emotion space is divided into four quadrants. Based on the emotion feature vector of the analysis text, the position in the two-dimensional emotion space is calculated, it is determined that the emotion state belongs to one of the above quadrants, and it is mapped to the corresponding flavor category.

5. The method of claim 4, wherein, The four quadrants are high pleasantness and high arousal, high pleasantness and low arousal, low pleasantness and high arousal, and low pleasantness and low arousal; and the emotion state corresponds to one of the four quadrants.

6. The method of claim 5, wherein, Determining that the emotion state belongs to one of the above quadrants and mapping to the corresponding flavor category comprises: reading the numerical values of the pleasantness and the arousal from the emotion feature vector; comparing the preset threshold values of the pleasantness and the arousal to obtain a quadrant index, retrieving the corresponding flavor category in a stored emotion-flavor mapping relationship table according to the quadrant index, and outputting the flavor category as an input parameter for subsequent recipe proportion generation. The flavor category includes fruit flavor, flower flavor, nut and vanilla flavor, and blended flavor.

7. The method of claim 6, wherein, Generating the recipe proportion of the flavor combination comprises: determining a target flavor category based on the emotion state, and selecting at least one specific flavor as a main flavor and at least one specific flavor as an auxiliary flavor in the target flavor category; selecting at least one blended flavor for overall flavor balance; determining the recipe proportion of the main flavor, the auxiliary flavor, and the blended flavor according to a preset proportion relationship, wherein the proportion of the main flavor is greater than the proportion of the auxiliary flavor, and the proportion of the auxiliary flavor is greater than the proportion of the blended flavor.

8. The method of claim 7, wherein, The recipe proportion of the flavor combination is generated, and further includes introducing a random disturbance within a preset range on the basis of the recipe proportion to increase the diversity of the flavor combination while maintaining the overall proportion relationship and outputting the recipe proportion of the flavor combination as an input parameter for the beverage dispensing device to perform dispensing.

9. The method of claim 8, wherein, The beverage dispensing device is controlled to perform dispensing according to the recipe proportion, and outputs a corresponding beverage, including, The recipe proportion is converted into an execution instruction of the beverage dispensing device, and the execution instruction includes a liquid output time or liquid output amount parameter corresponding to each flavor syrup pump; The base liquid supply unit is controlled to provide a liquid output amount of beverage base liquid, which is mixed with the flavor syrup in proportion, and the beverage base liquid includes at least one of coffee liquid, milk, or tea liquid; The stirring unit or flow channel mixing unit is synchronously controlled to fully mix and evenly mix the base liquid with each flavor syrup, and the dispensing operation is terminated when the beverage container reaches a preset capacity threshold, and the beverage is output.

10. A beverage intelligent blending system based on emotion recognition, employing the beverage intelligent blending method according to any one of claims 1 to 9. including, The collection module is configured to obtain a voice input of a user and transcribe the voice input into text through voice recognition; The emotion analysis module is configured to perform emotion analysis on the text using a trained language model, extract at least two emotion dimensions, and determine the emotion state of the user; The analysis text is input into the trained language model to obtain an emotion feature vector corresponding to the analysis text; the numerical values of at least two emotion dimensions are regressed or calculated based on the emotion feature vector; an emotion label and / or a quadrant category are generated according to the position of the emotion dimension in a preset emotion space to determine the emotion state; The recipe generation module is configured to map the emotion state to a corresponding flavor combination according to a preset emotion-flavor mapping mechanism, and generate a recipe proportion of the flavor combination, including reading the numerical values of the pleasure and the arousal from the emotion feature vector; comparing the numerical values of the pleasure and the arousal with preset threshold values to obtain a quadrant index; retrieving a corresponding flavor category from a stored emotion-flavor mapping relationship table according to the quadrant index, and outputting the flavor category as an input parameter for subsequent recipe proportion generation. The flavor category includes fruit aroma flavor, floral aroma flavor, nut and vanilla flavor, and blended flavor. Based on the emotion state, a target flavor category is determined, and at least one specific flavor is selected as a main flavor and at least one specific flavor is selected as an auxiliary flavor in the target flavor category; at least one blended flavor is selected for overall flavor balance; and the recipe proportion of the main flavor, the auxiliary flavor, and the blended flavor is determined according to a preset proportion relationship, wherein the proportion of the main flavor is greater than the proportion of the auxiliary flavor, and the proportion of the auxiliary flavor is greater than the proportion of the blended flavor. ​ The beverage dispensing control module is configured to control the beverage dispensing device to perform dispensing according to the recipe proportion, and output a corresponding beverage. The beverage dispensing control module includes: converting the recipe proportion into an execution instruction of the beverage dispensing device, wherein the execution instruction includes a liquid output time or a liquid output amount parameter corresponding to each flavor syrup pump; controlling a base liquid supply unit to provide a liquid output amount of a beverage base liquid, and proportionally mix the beverage base liquid with the flavor syrup, wherein the beverage base liquid includes at least one of coffee liquid, milk, or tea liquid; and synchronously controlling a stirring unit or a flow channel mixing unit to sufficiently and uniformly mix the base liquid with each flavor syrup, terminate the dispensing operation when a preset capacity threshold of a beverage container is reached, and output the beverage.

Images