Method and system for generating personalized printouts based on automatic speech recognition

Abstract

The present application relates to a kind of printing method and system based on voice automatic identification generation personalized printing output.The printing method includes the following steps: obtaining the real-time voice instruction of user;Extract the acoustic feature of user voice;According to the extracted acoustic feature, based on the preset acoustic feature grouping rule, judge the age group and / or gender group to which the user belongs;According to the age group and / or gender group to which the user belongs, determine the corresponding print style content;Using print style content, complete the personalized printing output of image and text.The present application determines the age group and / or gender group to which the user belongs by user voice, and automatically determines the corresponding print style content, which is beneficial to improve the pertinence of output content and user experience.

Description

Technical Field

[0001] This invention relates to the field of printing, and more specifically, to a printing method and system for generating personalized print output based on automatic voice recognition. Background Technology

[0002] Traditional printers typically offer only fixed print parameters / image output modes, requiring users to manually select paper type, print quality, filter effects, label styles, and retouching intensity via printer buttons or an app. This cumbersome process is not user-friendly, especially for children and the elderly. While voice interaction exists, it's limited to simple commands and lacks in-depth user analysis based on acoustic characteristics. Different genders and age groups have significantly different requirements for print parameters such as image style, color preferences, and text size (e.g., children prefer bright cartoons, while the elderly need high-contrast, large fonts). Traditional printers cannot automatically recognize user age and gender, nor can they adapt print parameters accordingly, failing to meet the diverse printing needs of different users. Furthermore, traditional printers cannot record user history, requiring repeated settings for each print job, resulting in inefficiency and an inability to dynamically adapt to changing user preferences. Summary of the Invention

[0003] To at least partially address the problems of the prior art, a first aspect of the present invention provides a printing method for generating personalized printouts based on automatic speech recognition, comprising the following steps: S11, obtain the user's real-time voice commands; S12, extract the acoustic features of the user's speech; S13. Based on the extracted acoustic features and the preset acoustic feature grouping rules, determine the user's age group and / or gender group. S14. Determine the corresponding print style content based on the user's age group and / or gender group; S15, using the aforementioned printing style content, complete the personalized printing output of the text and graphics.

[0004] Furthermore, the acoustic features include at least one of the fundamental frequency, the second formant, and the speech rate.

[0005] Furthermore, the print style content includes a print style template, and the determination of the print style template includes at least one of image filter selection, label style selection, and image editing parameter selection.

[0006] According to a specific embodiment of the present invention, step S14 retrieves the corresponding print style template from a preset style template library based on the user's age group and / or gender group; wherein, the style template library pre-stores filter candidate sets, label style candidate sets, and image editing parameter baseline values ​​for different groups.

[0007] Furthermore, the printing method also includes the following steps: S21, establish a behavior database for each registered user by user voiceprint or account, and record their most recent N historical print parameter selections; S22, perform statistics and calculations on the historical printing parameter selections, and generate a preference weight vector for each registered user.

[0008] Furthermore, the print style content includes a print style template, and step S14 includes: S141, Obtain the user's historical printing parameters, wherein the historical printing parameters include at least one of image filters, label styles, and image editing parameters; S142, Integrating user historical preferences, a corresponding print style template is generated. The generation of the print style template includes at least one of the following: Filter selection: Use the most frequently used filter in the registered user's history. If the most frequently used filter is not in the filter candidate set, the default filter will be used. Tag style selection: Use the most frequently used tags in the registered user's history. If the most frequently used tags are not in the tag style candidate set, the default tags will be used. Image retouching parameter calculation: The final image retouching parameters are obtained by weighting the average historical image retouching parameters of the registered user with the age group baseline value.

[0009] Furthermore, the printing method also includes the following steps: performing semantic analysis on the real-time voice command to extract keywords; determining whether there is a preset association rule between the keywords and the set label style; if so, in step S14, when determining the printing style template, the weight of the set label style is increased based on the user's historical preferences according to the following formula, and the label style with the highest weight is taken as the final output: W final =W history +Δ; where W final For the final weight after addition, W history The historical preference weights are generated based on the user's historical operations, and Δ is a preset fixed additive value; The association rule is the mapping relationship between the keyword and the preset tag style.

[0010] Furthermore, the printing style content includes image style templates, and determining the corresponding printing style content in step S14 includes selecting the corresponding image style template according to the user's age group and / or gender group.

[0011] A second aspect of the present invention discloses a printing system for generating personalized printouts based on automatic speech recognition, comprising: The voice acquisition module is used to acquire the user's real-time voice commands; The acoustic feature extraction module is used to preprocess the speech signal and extract the acoustic features of the user's speech; The recognition module is used to determine the user's age group and / or gender group based on preset acoustic feature grouping rules; The intelligent decision-making module is used to generate corresponding print style content based on the user's age group and / or gender. The print execution module is used to complete personalized printouts of text and images.

[0012] Furthermore, the printing system also includes: The user behavior database module is used to establish a corresponding behavior database for each user who completes registration through voiceprint or account, record and update the user's historical printing operation data, and generate a user preference weight vector. A style template library is used to store print style content corresponding to age groups and / or gender groups; wherein, the intelligent decision module is used to call or calculate the corresponding print style content from the style template library or the Internet according to the user's age group and / or gender group.

[0013] The technical solution of the present invention has the following beneficial effects: The printing method and system of the present invention determine the user's age group and / or gender group based on the acoustic characteristics of the user's voice, and automatically determine the corresponding printing style content based on the user group, so as to realize personalized printing output of text and graphics, greatly improve the relevance of the output content and user experience, and is especially convenient for special groups such as children and the elderly.

[0014] Furthermore, a behavioral database is established for each registered user, updating the preference weights of printing parameters such as filters, tag types, and retouching intensity in real time. This incorporates users' historical habits in subsequent printing processes, making the output more tailored to individual preferences. Moreover, a preferred embodiment of this invention integrates age and / or gender grouping, a pre-set style template library, user historical preferences, and semantic analysis to automatically generate the optimal printing strategy, balancing universality and personalization.

[0015] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the printing system framework in Example 1; Figure 2 This is a flowchart illustrating the printing method in Example 1; Figure 3 This is a schematic diagram of the printing system framework in Example 2; Figure 4 This is a flowchart illustrating the process of generating print style templates by integrating user history preferences in Example 2. Detailed Implementation

[0017] Example 1 Example 1 provides a printing method and system for generating personalized printouts based on automatic speech recognition. Please refer to [link to example]. Figure 1 The printing system includes: a voice acquisition module for acquiring real-time voice commands from the user; an acoustic feature extraction module for preprocessing the voice signal and extracting the acoustic features of the user's voice; a recognition module for determining the user's age group and / or gender group based on preset acoustic feature grouping rules; a style template library for storing printing style content corresponding to the age group and / or gender group; an intelligent decision-making module for calling or calculating corresponding printing style content from the style template library based on the user's age group and / or gender group; and a printing execution module for completing personalized printing output of graphics and text. Specifically, the graphics and text content generated and printed in this application includes at least one of the following: pure images, images and text, patterns and text, and pure text.

[0018] like Figure 2 As shown, the printing method for generating personalized printouts based on automatic speech recognition in Example 1 includes the following steps: S11, obtain the user's real-time voice commands; S12, extract the acoustic features of the user's speech; S13. Based on the extracted acoustic features and the preset acoustic feature grouping rules, determine the user's age group and / or gender group. S14. Determine the corresponding print style content based on the user's age group and / or gender group; S15, using the aforementioned printing style content, complete the personalized printing output of the text and graphics.

[0019] Specifically, step S11 acquires the user's real-time voice commands through a voice acquisition module (e.g., a microphone configured in the printer), and step S12 extracts the acoustic features of the user's voice through an acoustic feature extraction module. The acoustic features include at least one of the fundamental frequency, the second formant F2, and the speech rate. The fundamental frequency includes the fundamental frequency mean and fundamental frequency jitter. The method for extracting acoustic features can refer to existing technologies and will not be elaborated here.

[0020] In this invention, step S13 can refer to the example of grouping rules based on acoustic features in Table 1 below, and determine the user's age group and / or gender group through the recognition module. Further, step S13 can determine the grouping layer by layer in the order of first classifying gender and then age group, or it can directly combine age group and gender (for example, directly combine the extracted fundamental frequency mean, fundamental frequency jitter, second formant and speech rate to determine age group and gender).

[0021] Table 1: Examples of grouping rules based on acoustic features

[0022] For example, step S13 groups users by age group and / or gender based on hierarchical decision rules, including the following steps: S131, determine whether the user is a child.

[0023] The fundamental frequency mean is preferred as the criterion.

[0024] If the average base frequency is >280 Hz (the average base frequency for children is >300 Hz, with 280 Hz as a threshold margin), then the user is identified as a child. Optionally, step S131 can be combined with base frequency jitter for auxiliary confirmation when determining whether the user is a child. For example, the base frequency jitter can be required to be >0.4 to avoid misjudgment.

[0025] If the average fundamental frequency is ≤280 Hz, then the child is determined to be non-child, and the following step S132 is executed.

[0026] S132, determine if the user is an elderly person.

[0027] If the average base frequency is ≤280 Hz and the base frequency jitter is >0.5, the user is identified as an elderly user.

[0028] Optionally, step S132 further distinguishes gender: if the average fundamental frequency is <120 Hz, it is judged to be an elderly male; if the average fundamental frequency is ≥120 Hz and ≤280 Hz, it is judged to be an elderly female.

[0029] If the base frequency jitter is ≤0.5, then proceed to step S133 as follows.

[0030] S133, determine whether the user is a young person.

[0031] If the average base frequency is ≤280 Hz and the base frequency jitter is <0.3, it is judged to be a young user.

[0032] Optionally, step S133 further distinguishes gender: if the average fundamental frequency is >190 Hz (the average fundamental frequency of young women is 200-280 Hz, with a 190 Hz threshold reserved), it is judged as young women; if the average fundamental frequency is ≤190 Hz, it is judged as young men.

[0033] S134, determine middle-aged users.

[0034] If the average base frequency is ≤280 Hz and the base frequency jitter is ≥0.3 and <0.5, the user is judged to be middle-aged.

[0035] Optionally, step S134 further distinguishes gender: if the average fundamental frequency is >170 Hz, it is judged as a middle-aged woman; otherwise (i.e., the average fundamental frequency is ≤170 Hz), it is judged as a middle-aged man.

[0036] Furthermore, in steps S131 to S134 above, the second formant and speech rate of each gender and age group can be combined for auxiliary confirmation to improve the accuracy of the judgment.

[0037] In this invention, the style template library can be a local database of the printer, meaning the printer has a pre-built set of style templates for different age groups and / or genders; preferably, the style templates in the local database are added via firmware updates. The style template library can also be a cloud database; preferably, the style templates are updated regularly by obtaining image preferences for each age group and / or gender from the cloud, for example, by determining user preferences based on evaluation indicators such as the number of style template downloads and / or user ratings for each age group and / or gender, and then updating and adding style templates that match user preferences. Alternatively, the style template library can use both a local database and a cloud database simultaneously.

[0038] The print style content includes print style templates, which include image filters, label styles, and image editing parameters. The style template library pre-stores filter candidate sets, label style candidate sets, and image editing parameter baseline values ​​for different age groups and / or gender user groups. Step S14, determining the corresponding print style content, includes determining the corresponding print style template based on the user's age group and / or gender. Determining the print style template includes selecting at least one of the following: image filter selection, label style selection, and image editing parameter selection.

[0039] The filter candidate set includes several filter IDs, each filter corresponding to a set of color mapping parameters (such as a 3D LUT file). A filter refers to a set of digital image processing parameters that transform the overall tone and color style of an image, implemented through a preset color mapping matrix or lookup table. For example, filters include types such as Cartoon (suitable for children), High Saturation (suitable for young adults), Retro (suitable for middle-aged adults), and Black & White / Ink Painting (suitable for the elderly), as shown in Table 2 below. Filter Candidate Set F set Specify the default filter (e.g., {retro, high saturation, cartoon}) (e.g., the first filter in the candidate set).

[0040] Table 2: Examples of Common Filter Types

[0041] The candidate set of label styles includes several label type IDs. Label types include date labels, location labels, text quotes, and fun stickers. Each label type contains complete configuration parameters such as font, font size, color, border style, and position. Label style refers to the visual presentation configuration of text or graphic elements superimposed on a printed image, including font, color, decorative borders, and layout position. Common label types are shown in Table 3 below. Label styles / types are adapted to age groups and / or genders (e.g., rounded fonts + bright colors for children, bold fonts + large font sizes for the elderly). Candidate set of label styles T set (e.g., {date, quote, sticker}) Specify the default label.

[0042] Table 3: Examples of Tag Types

[0043] Image enhancement parameters refer to the quantitative control values ​​used to enhance images. These include general image enhancement parameters and scene-specific algorithm parameters, used to improve the visual effect of images. Image enhancement parameter baseline P base This is a set of default general image enhancement parameters (such as beautification intensity, contrast, saturation, brightness, and sharpness), as well as on / off flags for scene-specific image retouching algorithms (portrait skin smoothing, landscape sharpening, etc.). Preset baseline parameters for different age groups are shown in Table 4 below, and baseline parameters for different age groups plus gender groups are shown in Table 5 below.

[0044] Table 4: Examples of retouching parameters for different age groups

[0045] Table 5: Examples of retouching parameters for different age groups and genders

[0046] Image retouching parameter baseline values ​​explanation: Beauty enhancement intensity: Children and men are assigned lower values ​​to preserve a natural look; young women are assigned the highest values ​​to meet beautification needs; the elderly are assigned the lowest values ​​to preserve the authenticity of wrinkles.

[0047] Contrast ratio: moderately increased (1.1-1.3) for children, middle-aged and elderly, enhancing visual appeal; normal (1.0) for other age groups.

[0048] Saturation: Highest for children (1.3), to suit their preference for bright colors; moderately lower for older adults (0.9) to avoid glaring colors.

[0049] Brightness: Increased to 1.2 for older adults to help those with declining vision identify image details.

[0050] Sharpness: Increased moderately for older adults (1.2) to enhance image clarity; lowered slightly for children (0.8) to avoid overly sharp stimulation.

[0051] Optionally, the print style content also includes image style templates. Determining the corresponding print style content in step S14 includes selecting the corresponding image style template based on the user's age group and / or gender. Here, the image style template refers to generating images with different styles based on different age groups and / or gender groups. For example, children are matched with cartoon and crayon styles; young men and women with Japanese anime, fresh illustrations, and Instagram-style styles; middle-aged men and women with realistic photography, ink painting, and classic oil painting styles; and elderly young men and women with nostalgic retro and high-definition ink painting styles. For instance, if the system detects that a child wants to print a puppy, it will generate a cartoon-style puppy; if a middle-aged person wants to print a puppy, the system will generate a realistic-style puppy.

[0052] In addition, when generating image style templates, different styles of fonts and / or different font sizes can be generated according to different age groups. For example, children can generate rounded, thick-stroke, varying-size, cartoon-like fonts; young men and women can generate simple, clear, and uniform-line fonts; middle-aged men and women can generate Song, Kai, and Fangsong fonts; and elderly men and women can generate fonts with extremely thick strokes, high contrast, and no unnecessary decorations.

[0053] Alternatively, in another implementation, generating different styles of fonts and / or different font sizes based on different age groups can also be applied to simple text output, such as text labels generated for label printers or thermal printers.

[0054] In Example 1, step S14 involves grouping users according to their age group and / or gender. The intelligent decision-making module directly retrieves the corresponding print style content (e.g., print style templates) from a preset style template library. Step S15 involves the print execution module using the determined print style content to complete the personalized print output of the text and images.

[0055] Example 2 like Figure 3 As shown, the printing system in Embodiment 2 also includes a user behavior database module, which is used to establish a corresponding behavior database for each user who has registered through voiceprint or account, record and update the user's historical printing operation data, and generate a user preference weight vector.

[0056] Furthermore, in Example 2, the intelligent decision-making module calculates and generates a corresponding print style template based on the integration of users' historical preferences. That is, the printing method in Example 2 also includes the following steps: S21, establish a behavior database for each registered user based on the user's voiceprint or account, and record their most recent N historical print parameter selections; preferably, 5≤N≤10; S22, perform statistics and calculations on the historical printing parameter selections, and generate a preference weight vector for each registered user.

[0057] In this embodiment, the historical printing parameters include at least one of the following: preferred image filters (such as retro, black and white, high saturation, etc.), commonly used label types (such as date, location, text quotes, fun stickers), and image editing intensity (such as beauty level, contrast adjustment); preferably, the historical printing parameters also include paper type and printing quality (such as draft, standard, high quality).

[0058] The historical printing parameters selected by each registered user are statistically analyzed, and a preference weight vector is generated for each user and updated in real time. The calculation method is as follows: A fixed-length circular queue (such as an array) is maintained for each registered user, retaining the user's most recent N selections / operations (such as N=5 or 10), and preferences are calculated based on the data within the window. When calculating preferences, the frequency of each option in the queue is counted, and the option with the highest frequency is selected; the image editing values ​​are averaged. The advantages of this method are fixed memory usage (N records), simple calculation, and the ability to reflect recent trends.

[0059] When multiple options have the same highest frequency, the following methods can be used: Preset priority method, which predefines a default priority order for each option type (e.g., filter: cartoon > high saturation > retro > black and white), and selects the option with the highest priority when the frequencies are the same; or Recent priority method, which selects the option that was used most recently among the options with the same frequency, reflecting the user's latest preference.

[0060] In this embodiment, the preference weight vector describes the distribution of user preference for various options (e.g., filter A has a weight of 0.4, filter B has a weight of 0.3, and filter C has a weight of 0.3), and represents an intermediate statistical result. When making a specific printing decision, the system selects the option with the highest weight from the preference weight vector as the final output. If the weights are the same, the system decides based on the processing method where multiple options have the same frequency.

[0061] Taking user filter preference statistics (N=5) as an example, if the filter selection records of the user's most recent 5 operations are, in chronological order (from oldest to newest), retro, high saturation, retro, cartoon, retro, then the calculation and decision-making process of the filter preference weight vector is as follows: Step 1, frequency statistics: Retro style 3 times, High saturation style 1 time, Cartoon style 1 time; Step 2, calculate the preference weight vector: weight = frequency of each option / total frequency, retro weight = 3 / 5 = 0.6, high saturation weight = 1 / 5 = 0.2, cartoon weight = 1 / 5 = 0.2, other filter weights = 0; Step 3, output the preference weight vector: [Retro: 0.6, High Saturation: 0.2, Cartoon: 0.2]; Step 4, Decision Output (select the one with the highest weight): Retro has the highest weight of 0.6, so the "Retro" filter is selected for this print.

[0062] If the highest weights are tied, a conflict resolution rule (such as the nearest priority method) is used. For example, the four most recent operations are, in chronological order (from oldest to newest): Retro, High Saturation, Retro, High Saturation. Retro: 2 times (weight 0.5); High Saturation: 2 times (weight 0.5); others: 0 times. The most recent operation is "High Saturation", so "High Saturation" is selected.

[0063] In Example 2, step S14 generates a corresponding print style template based on the integration of user historical preferences, including: S141, Obtain the user's historical printing parameters, wherein the historical printing parameters include at least one of image filters, label styles, and image editing parameters; S142, integrates user's historical preferences to generate corresponding print style templates.

[0064] Specifically, the generation of the print style template in step S142 includes: Filter selection: Use the user's most frequently used filter from their history. If the most frequently used filter is not in the filter candidate set, use the default filter. For example, select the user's most frequently used filter from their history (f). user If f user Belongs to F set Then the filter F=f user Otherwise, use the default filter. Tag style selection: Use the user's most frequently used tags from their registration history. If the most frequently used tag is not in the tag style candidate set, the default tag is used. For example, select the user's most frequently used tag t. user If t user Belongs to T set Then T=t user Otherwise, use the default tag. Image retouching parameter calculation: The final image retouching parameters are obtained by weighting the average historical image retouching parameters of the registered user with the age group baseline value.

[0065] Image editing parameter calculation method: Take the user's historical average value P user (e.g., beauty filters, contrast); if the user has sufficient history (e.g., at least 1 time), a weighted average is used, with the specific steps as follows: 1. Calculate the user's historical average value P. user The arithmetic mean of the image editing parameters (beauty intensity, contrast, saturation, brightness, and sharpness) in the user's most recent N operation records is calculated to obtain the user's personal preference mean.

[0066] 2. Weighted fusion: This involves combining the user's historical average value P. user Compared with the age group baseline value P base We perform a weighted average to obtain the final retouching parameter P, where P = α·P base +(1-α)·P user The range of values ​​for the image editing parameter P can be found in Table 6 below.

[0067] Table 6: Examples of Image Editing Parameters and Their Value Ranges

[0068] In this invention, the weight α can be dynamically adjusted based on the amount of user historical data. If the number of historical occurrences is less than a set number n, the weight α is increased; if the number of historical occurrences is greater than or equal to the set number n, the weight α is decreased. For example, if the number of historical occurrences n < 3, then α = 0.6 (more trust age group benchmark); if the number of historical occurrences n ≥ 3, then α = 0.3 (more trust user historical preferences).

[0069] If there is no historical data (first time using), directly use P=P base .

[0070] Specific calculation example 1 (taking beautification intensity as an example): User's last 5 historical records: 0.4, 0.5, 0.4, 0.6, 0.5; P user =(0.4+0.5+0.4+0.6+0.5) / 5=0.48 (arithmetic mean); Take Pbase =0.3 (children's baseline), α=0.3; P=0.3×0.3+0.7×0.48=0.09+0.336=0.426.

[0071] Among them, P base The values ​​of (age group retouching parameters benchmark) are based on the following comprehensive considerations: differences in visual perception and aesthetic preferences among users of different age groups; and the collection of preference ratings for retouching effects from users of different age groups through small-scale user surveys.

[0072] Specific calculation example 2: When an elderly female user uses the system for the first time, the system directly uses P. base The settings are: Beauty Intensity = 0.1, Contrast = 1.3, Saturation = 0.9, Brightness = 1.2, Sharpness = 1.2. After processing, the image contrast is improved, the brightness is increased, the color saturation is slightly reduced, and the skin smoothing is very minor. The final output is clear, easy to distinguish, and retains a realistic effect.

[0073] Furthermore, during image editing, the system automatically identifies the image content (such as determining whether it is a human portrait through face detection), activates the corresponding scene algorithm (such as portrait skin smoothing and landscape sharpening, see Table 7 for details), and makes final adjustments based on age group baseline parameters (such as improving contrast and brightness for elderly portraits).

[0074] Table 7: Examples of Scene Algorithms

[0075] For example, the decision-making process for enabling the corresponding scenario algorithm is as follows: Input: Original image + age group / gender Step 1, Image Content Recognition / Face Detection: If a face is detected, it is marked as a portrait scene; if no face is detected, edge analysis is performed to determine whether it is a landscape / text / default.

[0076] Step 2, select the scene processing algorithm: If it is a portrait scene, enable the skin smoothing algorithm (bilateral filtering); if it is a landscape scene, enable the sharpening algorithm; if it is another scene, only adjust the general parameters.

[0077] Step 3: Determine the retouching parameters based on age group and / or gender.

[0078] For example, look up the following parameters in a table: beauty intensity, contrast, saturation, brightness, and sharpness; Portrait scenes: Skin smoothing intensity = Beauty enhancement intensity; Landscape scene: Sharpness = lookup table value; General adjustments: Contrast, saturation, and brightness should be adjusted according to the table values; Step 4: Output the processed image using the determined print style template.

[0079] Furthermore, determining the print style template also includes selecting the paper type and print quality. The paper type and print quality can be the user's most frequently used selections; otherwise, the default is "Standard".

[0080] In other embodiments of the present invention, the printing system may not include a style template library. In step S14, the intelligent decision-making module can retrieve or calculate corresponding printing style content from the Internet based on the user's age group and / or gender. Specifically, "retrieving or calculating corresponding printing style content from the Internet" means that it is not necessary to establish a cloud database, but rather to directly search for and retrieve the corresponding group of printing style content from the Internet based on the user's age group and / or gender.

[0081] Example 3 Example 3 further adds a semantic analysis module based on the previous examples and considers image content tags. If there are preset association rules between image content tags and certain tag styles (e.g., "birthday" tends to use the "date" tag), the corresponding tag styles can be weighted based on user preferences. Here, image content tags are defined as keywords extracted from user voice commands, describing the content the user wants to print, and are used for image retrieval and tag style-assisted decision-making.

[0082] The printing method in this embodiment further includes the following steps: the semantic analysis module performs semantic analysis on the real-time voice command and extracts image content tags / keywords; if the keywords and the set tag style have a preset association rule, then in step S14, the weight of the set tag style is increased when determining the printing style template.

[0083] In this embodiment, the extraction of image content tags / keywords can be carried out using the following method: after converting the user's voice into text, stop word filtering and noun extraction are performed to obtain image content tags / keywords (such as "puppy", "birthday cake", "scenery").

[0084] The system's association rule library pre-defines association rules between image content tags and tag styles. These association rules represent the preset mapping relationship between keywords and set tag styles. For example, "birthday," "cake," and "party" are associated with / mapped to "date tags"; "travel" and "attractions" are associated with / mapped to "location tags"; "quotes" and "sayings" are associated with / mapped to "text quotes"; and "thank you" and "blessings" are associated with / mapped to "fun stickers."

[0085] Based on the weights of users' historical preferences, if the current image content tags / keywords match the association rules, the corresponding tag style is weighted (e.g., weight +0.2 or multiplied by 1.2), and the tag style with the highest weight is taken as the final output. If the rules are not matched, the default selection or the result of users' historical preferences is used directly.

[0086] In this embodiment, the specific steps for adding image content tag / keyword related tag style weights based on user preferences include: 1. Obtain the user preference weight vector: Obtain the user's historical preference weight for each tag style (based on the statistics of the most recent N operations, weight = number of occurrences / N).

[0087] 2. Apply weighted bonus: Determine whether there is a preset association rule between the extracted keywords and the set tag style. If the current image content tag matches the association rule, then add a bonus to the associated tag style.

[0088] For example, the weight fusion uses a linear addition strategy. If the currently extracted image content tag matches a preset association rule, then the style of the associated target tag is weighted, and the calculation formula is as follows: W final =W history +Δ Among them, W final For the final weight after addition, W history The weights are historical preference weights generated based on users' historical operations (e.g., frequency weights derived from circular queue statistics). Δ is a preset fixed additive value; in this embodiment, Δ is set to 0.2. The weights are not normalized after additive processing and are used directly for comparison.

[0089] For example: User history preferences: Fun stickers weighted 0.6, text quotes weighted 0.4, date tags weighted 0.0; User's voice message: "Print birthday cake" → hits "birthday" → strongly associates with "date tag"; Weighting boost (+0.2): The date tag weight becomes 0.2; Final weights: Fun stickers 0.6, text quotes 0.4, date tags 0.2 → Select "Fun stickers".

[0090] 3. Decision Output: Select the label style with the highest weight after the bonus as the label type for this printing. If multiple labels have the same weight, use the conflict resolution rules mentioned above (such as the nearest-to-last method).

[0091] 4. Output the final strategy: Pass the determined filter ID, label style ID, image editing parameters and paper type to the print execution module to generate the image and print it.

[0092] Furthermore, the printing method of this embodiment also includes the following steps: searching in a preset image library and / or style template library according to the image content tags / keywords to match the target image to be printed; in step S15, the target image is printed using the determined printing style content as the image to be printed.

[0093] Example

[0094] I. Scene Setting User: Elderly woman, first time using.

[0095] II. First-time use (no prior history preference) Step 1: Voice Acquisition and Age Recognition The user said, "Print a photo."

[0096] The system extracted the following acoustic features: fundamental frequency 165Hz, jitter 0.65, speech rate 2.8 syllables / second; Judgment rules: Base frequency 170-300Hz with jitter >0.5 and speech rate <3.0 → elderly woman; Step 2: Obtain age group benchmarks Filter candidate set: {Retro, Black and White, Ink Painting} → Default is "Retro" Tag candidate set: {Date tags, text quotes, fun stickers} → Default is "Date tags" Baseline retouching parameters: Beauty 0.1, Contrast 1.3, Saturation 0.9, Brightness 1.2, Sharpness 1.2.

[0097] Step 3: Intelligent Decision Making (No History) Filter = Retro, Tag = Date Tag, Editing Parameters = Baseline Value.

[0098] Step 4: Image Editing and Printing Image processing: Sharpening (1.2) + Contrast (1.3) + Brightness (1.2) + Saturation (0.9); Output: Retro-style landscape photo with a date label overlaid in the lower right corner.

[0099] Step 5: Record preferences User behavior queue: [Retro, Date tag].

[0100] III. Second Use (Integrating Historical Preferences) The user said, "Print a birthday cake."

[0101] Step 1: Extract content tags Speech recognition → Keyword "birthday cake".

[0102] Step 2: Obtain user preferences (based on the most recent N=2 records) Filter preference: Retro weight 1.0 → Select "Retro" (in the candidate set); Tag preference: Date tag weight 1.0 → Temporarily select "Date tag".

[0103] Step 3: Content Tag Weighting The "birthday cake" match the association rule → strongly associated with "date tag", and the weight of "date tag" is increased by 0.2 → still the highest, so "date tag" is ultimately selected.

[0104] Step 4: Image editing parameter blending If the number of historical occurrences is less than 3, α = 0.6: P = 0.6 × baseline + 0.4 × user history.

[0105] User history = baseline (first use) → still the baseline value.

[0106] Step 5: Print out Birthday cake picture + date label.

[0107] Step 6: Update preferences Queue update: [Retro, Date Tag, Retro, Date Tag].

[0108] In summary, this invention automatically identifies gender and age group based on user voice and automatically calls the corresponding style template library to achieve age-appropriate adaptation of filters, tag styles, and image editing parameters, greatly improving the relevance of the output content and user experience, and is especially convenient for children and the elderly.

[0109] Furthermore, a behavioral database is established for each registered user, using a sliding window to record the most recent N operations. Preference weights for filters, tag types, and retouching intensity are updated in real time. These preferences are then incorporated into subsequent printing, making the output more tailored to individual preferences. The algorithm is lightweight and suitable for embedded devices. In addition, an intelligent decision-making module combining priority and weighted averaging, integrating voiceprint age groups, user history preferences, a pre-set template library, and image content tags, automatically generates optimal printing strategies. For example, it prioritizes commonly used filters based on age group criteria and applies weighted smoothing to retouching parameters, balancing universality and personalization.

[0110] Although the present invention has been described above through embodiments, it should be understood that the above embodiments are only used to exemplarily describe possible implementations of the present invention and should not be construed as limiting the scope of protection of the present invention. That is, any substitutions or changes made by those skilled in the art in accordance with the present invention should also be covered by the scope of protection of the claims of the present invention.

Claims

1. A printing method for generating personalized printouts based on automatic speech recognition, characterized in that... Includes the following steps: S11, obtain the user's real-time voice commands; S12, extract the acoustic features of the user's speech; S13. Based on the extracted acoustic features and the preset acoustic feature grouping rules, determine the user's age group and / or gender group. S14. Determine the corresponding print style content based on the user's age group and / or gender group; S15, using the aforementioned printing style content, complete the personalized printing output of the text and graphics.

2. The printing method according to claim 1, characterized in that: The acoustic features include at least one of the fundamental frequency, the second formant, and the speech rate.

3. The printing method according to claim 1, characterized in that: The print style content includes print style templates, and the determination of the print style templates includes at least one of image filter selection, label style selection, and image editing parameter selection.

4. The printing method according to claim 3, characterized in that: Step S14: Based on the user's age group and / or gender, retrieve the corresponding print style template from the preset style template library; wherein, the style template library pre-stores filter candidate sets, label style candidate sets, and image editing parameter baseline values ​​for different groups.

5. The printing method according to claim 1, characterized in that... It also includes the following steps: S21, establish a behavior database for each registered user by user voiceprint or account, and record their most recent N historical print parameter selections; S22, perform statistics and calculations on the historical printing parameter selections, and generate a preference weight vector for each registered user.

6. The printing method according to claim 5, characterized in that, The print style content includes a print style template, and step S14 includes: S141, Obtain the user's historical printing parameters, wherein the historical printing parameters include at least one of image filters, label styles, and image editing parameters; S142, Integrating user historical preferences, a corresponding print style template is generated. The generation of the print style template includes at least one of the following: Filter selection: Use the most frequently used filter in the registered user's history. If the most frequently used filter is not in the filter candidate set, the default filter will be used. Tag style selection: Use the most frequently used tags in the registered user's history. If the most frequently used tags are not in the tag style candidate set, the default tags will be used. Image retouching parameter calculation: The final image retouching parameters are obtained by weighting the average historical image retouching parameters of the registered user with the age group baseline value.

7. The printing method according to claim 6, characterized in that... It also includes the following steps: Semantic analysis is performed on the real-time voice commands to extract keywords; Determine whether there is a preset association rule between the keyword and the set label style. If there is, in step S14, when determining the print style template, the weight of the set label style is increased based on the user's historical preferences according to the following formula, and the label style with the highest weight is taken as the final output. W final =W history +Δ Among them, W final For the final weight after addition, W history The historical preference weights are generated based on the user's historical operations, and Δ is a preset fixed additive value; The association rule is the mapping relationship between the keyword and the preset tag style.

8. The printing method according to any one of claims 1 to 7, characterized in that: The printing style content includes image style templates. In step S14, determining the corresponding printing style content includes selecting the corresponding image style template based on the user's age group and / or gender.

9. A printing system for generating personalized printouts based on automatic speech recognition, characterized in that... include: The voice acquisition module is used to acquire the user's real-time voice commands; The acoustic feature extraction module is used to preprocess the speech signal and extract the acoustic features of the user's speech; The recognition module is used to determine the user's age group and / or gender group based on preset acoustic feature grouping rules; The intelligent decision-making module is used to generate corresponding print style content based on the user's age group and / or gender. The print execution module is used to complete personalized printouts of text and images.

10. The printing system according to claim 9, characterized in that... Also includes: The user behavior database module is used to establish a corresponding behavior database for each user who completes registration through voiceprint or account, record and update the user's historical printing operation data, and generate a user preference weight vector. A style template library is used to store print style content corresponding to age groups and / or gender groups; wherein, the intelligent decision module is used to call or calculate the corresponding print style content from the style template library or the Internet according to the user's age group and / or gender group.

Images