Loader vehicle-mounted voice interaction system and use method thereof

Abstract

The invention relates to a loader vehicle-mounted voice interaction system and a use method. The system comprises a noise acquisition and processing module, a voice recognition module, a natural language processing and dialogue management module and a text-to-voice module. The noise acquisition and processing module processes low-frequency, intermediate-frequency and high-frequency noise by adopting Kalman filtering, Wiener filtering and frequency spectrum subtraction, so that the definition of voice signals is improved; the voice recognition module recognizes a voice instruction through a deep learning model; the natural language processing module analyzes and executes the instruction in combination with the operation state of the loader; and the text-to-voice module dynamically adjusts the feedback volume and speed according to the noise environment. The voice recognition accuracy is improved, the man-machine interaction experience is optimized, and the method has the advantages of being easy to operate, safe, reliable, high in intelligent degree and the like.

Description

Technical Field

[0001] The invention relates to a vehicle-mounted voice interaction technology, in particular to a loader vehicle-mounted voice interaction system and a use method thereof. Background Art

[0002] With the development of automation technology, heavy engineering machinery and equipment such as loaders are increasingly used in high-noise environments such as construction and mining. When operating these mechanical equipment, operators usually need to perform a large number of complex operations and frequently rely on manual control. However, the traditional operation mode not only consumes a lot of manpower and energy, but is also prone to low operating efficiency and even misoperation due to factors such as environmental noise and operational complexity. In addition, the noise sources of modern engineering machinery and equipment in the working environment are complex and diverse, such as engine noise, mechanical vibration, wind noise, etc., which makes the existing voice control system easily interfered when recognizing voice commands, resulting in a decrease in recognition accuracy.

[0003] In the existing technology, there has been some progress in the development of vehicle-mounted voice interaction systems, especially the widespread application of speech recognition (ASR) and natural language processing (NLP) technologies. However, traditional speech recognition systems are often unable to adapt to the special working conditions of loaders. Especially in high-noise environments, existing noise reduction methods are difficult to achieve real-time and dynamic noise filtering, resulting in the inability to effectively guarantee the accuracy and stability of speech recognition. At the same time, existing natural language processing technology is mainly used in daily conversation scenarios and cannot adapt to special instructions and operating requirements in loader operations.

[0004] To this end, it is urgent to develop an on-board voice interaction system that can cope with complex working conditions and high-noise environments, so as to improve the intelligence and safety of loader operations and optimize the operator's working experience. Summary of the invention

[0005] The present invention provides a loader vehicle-mounted voice interaction system and a method for using the system, aiming to solve the problems of inaccurate voice recognition in a high-noise environment, complex operator interaction, and insufficient intelligence level.

[0006] The present invention provides a loader vehicle-mounted voice interaction system, the system comprises a noise collection and processing module, a voice signal collection and recognition module, a natural language processing and dialogue management module, a text-to-speech module and a personalized voice synthesis module, wherein: The noise collection and processing module collects noise data in the environment in real time through noise sensors installed at multiple key locations of the loader, and uses Kalman filtering, Wiener filtering and spectrum subtraction algorithms to process noise in different frequency bands, ensuring that the voice signal is separated from the noise and transmitted to the voice recognition module in a clear state; The voice signal acquisition and recognition module uses a deep learning model to recognize the noise-reduced voice signal. The system uses voice activation detection technology to ensure the extraction of valid voice signals and uses ASR voice recognition to convert them into executable text instructions. The natural language processing and dialogue management module analyzes the recognized voice commands based on NLP technology, and dynamically adjusts the task priority based on the current working conditions of the loader, generates commands and executes the corresponding tasks; The text-to-speech module dynamically adjusts the volume, speed and tone of voice prompts according to the real-time noise environment of the loader, ensuring that the operator can clearly receive feedback information in different noise environments.

[0007] Furthermore, the noise collection and processing module uses the Kalman filter algorithm to process low-frequency noise, the Wiener filter to process intermediate-frequency noise, and the spectrum subtraction to process high-frequency noise.

[0008] Furthermore, the voice signal acquisition and recognition module utilizes voice activation detection technology to distinguish effective voice signals from background noise signals to improve the accuracy and efficiency of voice command recognition.

[0009] Furthermore, the natural language processing and dialogue management module monitors the loader's operating status, load conditions, fuel consumption and other data in real time, intelligently analyzes the current working conditions and prioritizes the execution of key task instructions.

[0010] Furthermore, the text-to-speech module dynamically adjusts output parameters according to the real-time noise environment, increasing the volume and speech rate when the loader is in a high-noise state and reducing the volume in a quiet environment to ensure the clarity and adaptability of the voice feedback.

[0011] Furthermore, it also includes a personalized speech synthesis module, which allows the operator to set different voice prompt styles, including a concise mode and a detailed mode, and automatically saves the operator's preference settings to maintain a consistent operating experience during long-term use.

[0012] The present invention also provides a method for using the loader vehicle-mounted voice interaction system, comprising the steps of: The noise is collected through sensors and divided into low-frequency noise, medium-frequency noise and high-frequency noise; Adaptive Kalman filter algorithm is used to process the low-frequency noise of the loader voice signal; The intermediate frequency noise of the loader speech signal is processed by Wiener filtering; The high-frequency noise of the loader's voice signal is processed by spectrum subtraction; Extract and recognize the de-noised language signal to generate text instructions; The recognized text instructions are matched with the internal instruction library of the loader. If the matching degree between the recognized text and the known instructions is higher than the threshold, the instructions will be verified. Parse verified instructions using pre-trained NLP models; Get the working status of the loader and determine the priority of the instruction; The task execution order is dynamically adjusted according to the priority of the instructions and passed to the execution module for execution.

[0013] Furthermore, in the step of dynamically adjusting the task execution order according to the priority of the instructions and passing them to the execution module for execution, it also includes: after adjusting the task order, generating corresponding feedback information to inform the operator of the execution status of the current task.

[0014] Furthermore, the feedback information is converted from text to voice playback, including: Detect the noise level in the current environment; Automatically adjust the output volume and speech speed according to the real-time noise level; The parsed text information generates voice prompts through the TTS engine; The generated voice feedback is played to the operator via the vehicle audio system. Beneficial Effects

[0015] By integrating multiple advanced technologies such as automatic speech recognition (ASR), natural language processing (NLP), and text-to-speech (TTS), the present invention can significantly improve the automation and intelligence level of loader operation while optimizing the operator's experience.

[0016] The system achieves accurate recognition and execution of loader voice commands through close collaboration of multiple modules, ensuring that the system can work efficiently and stably in complex environments. The following are the main modules and working principles of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is an architecture diagram of the loader vehicle-mounted voice interaction system according to an embodiment of the present invention.

[0018] Figure 2 The present invention is a schematic diagram of the use process of the loader vehicle-mounted voice interaction system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The present invention is further described below in conjunction with specific embodiments and drawings.

[0020] See also Figure 1 and Figure 2This embodiment provides a loader vehicle-mounted voice interaction system, including: a noise collection and processing module, a voice signal collection and recognition module, a natural language processing and dialogue management module, a text-to-speech module and a personalized speech synthesis module, wherein: The noise collection and processing module collects noise data in the environment in real time through noise sensors installed at multiple key locations of the loader, and uses Kalman filtering, Wiener filtering and spectrum subtraction algorithms to process noise in different frequency bands, ensuring that the voice signal is separated from the noise and transmitted to the voice recognition module in a clear state; The voice signal acquisition and recognition module uses a deep learning model to recognize the noise-reduced voice signal. The system uses voice activation detection technology to ensure the extraction of valid voice signals and uses ASR voice recognition to convert them into executable text instructions. The natural language processing and dialogue management module analyzes the recognized voice commands based on NLP technology, and dynamically adjusts the task priority based on the current working conditions of the loader, generates commands and executes the corresponding tasks; The text-to-speech module dynamically adjusts the volume, speed and tone of voice prompts according to the real-time noise environment of the loader, ensuring that the operator can clearly receive feedback information in different noise environments.

[0021] Furthermore, the noise collection and processing module uses the Kalman filter algorithm to process low-frequency noise, the Wiener filter to process intermediate-frequency noise, and the spectrum subtraction to process high-frequency noise.

[0022] Furthermore, the voice signal acquisition and recognition module utilizes voice activation detection technology to distinguish effective voice signals from background noise signals to improve the accuracy and efficiency of voice command recognition.

[0023] Furthermore, the natural language processing and dialogue management module monitors the loader's operating status, load conditions, fuel consumption and other data in real time, intelligently analyzes the current working conditions and prioritizes the execution of key task instructions.

[0024] Furthermore, the text-to-speech module dynamically adjusts output parameters according to the real-time noise environment, increasing the volume and speech rate when the loader is in a high-noise state and reducing the volume in a quiet environment to ensure the clarity and adaptability of the voice feedback.

[0025] Furthermore, it also includes a personalized speech synthesis module, which allows the operator to set different voice prompt styles, including a concise mode and a detailed mode, and automatically saves the operator's preference settings to maintain a consistent operating experience during long-term use.

[0026] The present invention also provides a method for using the loader vehicle-mounted voice interaction system, comprising the steps of: The noise is collected through sensors and divided into low-frequency noise, medium-frequency noise and high-frequency noise; Adaptive Kalman filter algorithm is used to process the low-frequency noise of the loader voice signal; The intermediate frequency noise of the loader speech signal is processed by Wiener filtering; The high-frequency noise of the loader's voice signal is processed by spectrum subtraction; Extract and recognize the de-noised language signal to generate text instructions; The recognized text instructions are matched with the internal instruction library of the loader. If the matching degree between the recognized text and the known instructions is higher than the threshold, the instructions will be verified. Parse verified instructions using pre-trained NLP models; Get the working status of the loader and determine the priority of the instruction; The task execution order is dynamically adjusted according to the priority of the instructions and passed to the execution module for execution.

[0027] Furthermore, in the step of dynamically adjusting the task execution order according to the priority of the instructions and passing them to the execution module for execution, it also includes: after adjusting the task order, generating corresponding feedback information to inform the operator of the execution status of the current task.

[0028] Furthermore, the feedback information is converted from text to voice playback, including: Detect the noise level in the current environment; Automatically adjust the output volume and speech speed according to the real-time noise level; The parsed text information generates voice prompts through the TTS engine; The generated voice feedback is played to the operator via the vehicle audio system.

[0029] The detailed operation principle of the above system is as follows: 1. Noise collection and processing module Implementation principle: In the working environment of the loader, noise is the main interference factor for speech recognition. Noise sources include engine noise, mechanical vibration, wind noise, etc. These noises usually have multiple frequency components, which not only cover the frequency band of the speech signal, but also interfere with the accuracy of the speech recognition system. In order to overcome this problem, the noise processing module of the present invention collects noise data in real time by installing noise sensors at multiple locations of the loader, and uses multiple noise processing algorithms (such as Kalman filtering, Wiener filtering and spectral subtraction) to process the noise. By processing different noise sources in different frequency bands, the clarity of the system's speech signal can be significantly improved.

[0030] content: The main function of this module is to ensure that the speech signal is separated from the noise and passed to the speech recognition module with sufficient clarity. By installing multiple noise sensors and processing noise data in real time, the system can effectively cope with the complex noise environment of the loader. The module uses Kalman filtering to process low-frequency noise, Wiener filtering to process medium-frequency noise, and spectral subtraction to process high-frequency noise, ensuring the recognizability of speech signals under the interference of noise of different frequencies.

[0031] step: Noise collection: The system installs multi-channel noise sensors at key locations of the loader (key locations include the cab, around the engine, and outside the vehicle body). Each sensor is responsible for collecting noise signals in a specific area, ensuring that noise changes at different locations can be monitored in real time.

[0032] These sensors continuously record noise in the environment, including mechanical vibrations (such as hydraulic systems), engine noise (such as diesel engines), wind noise, etc. Each type of noise has different frequency characteristics and therefore needs to be processed separately.

[0033] Low frequency noise processing: For low-frequency noise (such as mechanical vibration), the system uses an adaptive Kalman filter algorithm to process it. The Kalman filter can model low-frequency noise and dynamically adjust its parameters in real time to filter out low-frequency interference while retaining the voice signal.

[0034] For example, the mechanical vibration of hydraulic equipment during operation will generate low-frequency noise. The Kalman filter can dynamically estimate the noise and suppress it to avoid affecting the voice signal.

[0035] IF noise processing: Intermediate frequency noise (such as engine noise) is processed by Wiener filtering. Wiener filtering is a noise processing method based on the frequency domain, which can optimize the signal-to-noise ratio of the signal in the frequency domain. The system analyzes the noise spectrum, finds the noise components, and compresses and filters them.

[0036] The intermediate frequency noise emitted by the engine when it is working can easily interfere with the voice signal. The Wiener filter can effectively remove these noise components and ensure the accuracy of speech recognition.

[0037] High frequency noise processing: High-frequency noise (such as wind noise) is processed by spectral subtraction technology. Spectral subtraction can identify high-frequency noise components from the spectrum of the speech signal and subtract these noises from the spectrum to reduce the impact of high-frequency noise.

[0038] For example, when a loader moves at high speed, wind noise will generate high-frequency noise, and spectrum subtraction can effectively reduce the interference of this noise on the voice signal.

[0039] Signal output: After the above processing, the signal has removed the noise in each frequency band and retained a relatively pure voice signal. The system transmits the denoised voice signal to the voice recognition module for subsequent processing.

[0040] 2. Voice signal acquisition and recognition module Implementation principle: After the noise is processed, the system needs to extract valid voice signals and recognize them. In order to ensure accurate extraction of voice from the noise-processed signal, the present invention uses advanced voice recognition technology. The voice recognition engine is based on a deep learning model and can accurately recognize voice commands in complex environments and convert them into executable text commands.

[0041] Voice activation detection improvements: By combining voice activation detection with the working status of the loader, the system can adaptively adjust the sensitivity of voice activation. For example, when the loader is working under heavy load and the background noise is large, the system will increase the activation threshold to avoid false triggering of recognition by noise; while in a stationary state, the system will lower the threshold to ensure that weaker voice signals are captured.

[0042] Customized training of deep learning models: We collected and trained voice samples in the working scene of the loader, including the common commands and speech speed of operators in noisy and quiet environments. At the same time, we sampled various background noises and added them to the training data of the model, so that the model can adapt to the voice input in the high noise and complex environment of the loader.

[0043] Aiming at the complex noise environment of the loader (including engine noise, mechanical vibration, wind noise, etc.), we have made in-depth improvements based on the existing technology, which are mainly reflected in the following aspects: Multi-channel noise acquisition and spatial filtering: By installing multiple noise sensors at multiple key locations of the loader (such as the cab, near the engine, outside the vehicle body, etc.), the system can distinguish noise sources from different locations through spatial filtering technology, thereby processing noise more accurately. This technology combines the spatial characteristics of noise and can dynamically weight noise according to location to ensure the accuracy of noise reduction.

[0044] Semantic analysis combined with real-time working conditions: The system monitors the loader's operating status in real time, such as current load, fuel consumption, working mode and other data. Through these working condition information, combined with voice commands, the system can more intelligently understand and execute the operator's semantics. For example, when the operator issues a "stop operation" command, the system will determine the current status of the loader (whether it is moving, whether it is fully loaded, etc.), and thus determine the safest and most effective way to stop the machine.

[0045] Dynamic mapping and priority sorting of instruction libraries: The system not only relies on a fixed command library, but also executes voice commands by adjusting task priorities in real time. For example, commands such as emergency braking or equipment fault handling will be executed first, while non-urgent tasks will be delayed. This mechanism dynamically maps commands through NLP, allowing the system to execute efficiently according to task priorities.

[0046] content: The core of this module is to use the de-noised voice signal to perform the voice recognition task. Through the training and optimization of the deep learning model, the system can cope with different voice signal variations and inconsistencies, ensuring that voice commands can be accurately recognized and parsed in a high-noise environment.

[0047] step: Signal Input: The denoised signal is passed from the noise processing module to the speech recognition module. At this point, the signal has been processed by multi-band noise, ensuring that the speech component in the signal is enhanced and the noise component is effectively suppressed.

[0048] Speech signal extraction: The system detects valid speech components in the input signal to ensure that the extracted signal is the part with sufficient speech information. This stage involves further processing of the signal, such as voice activation detection (VAD), to distinguish speech signals from non-speech signals (such as short pauses, noise, etc.).

[0049] Voice Recognition: The ASR speech recognition engine is used to perform real-time recognition of the extracted speech signals. Whisper is a Transformer-based model that can process speech input in various contexts and recognize text instructions.

[0050] The recognition result is a text instruction generated by the system for further processing by subsequent modules.

[0051] Instruction verification: The system matches the recognized text instructions with the loader's internal instruction library to ensure that the operator's instructions are accurately interpreted. If the recognized text matches the known instructions, the instructions will be verified.

[0052] In some cases, the system may perform a secondary verification of the recognition results, especially in noisy conditions, and the system may use redundant mechanisms to ensure the accuracy of the instructions.

[0053] Output instructions: The verified instructions will be passed to the next module, the natural language processing module, for parsing to ensure the effective execution of the voice instructions.

[0054] 3. Natural Language Processing and Dialogue Management Module Implementation principle: After the system recognizes the voice command, it must further analyze the command and perform the corresponding task according to the current state of the loader. This module uses natural language processing (NLP) technology to understand the operator's intentions and prioritize task execution based on actual working conditions to ensure safe and efficient completion of operations.

[0055] content: The natural language processing and dialogue management module is responsible for semantic analysis of the operator's instructions. The system understands the operator's commands through a locally deployed and fine-tuned language model, dynamically adjusts the execution order of the instructions based on the loader's current state and task requirements, generates appropriate feedback, and executes the corresponding tasks.

[0056] step: Voice command analysis: The system uses a pre-trained NLP model to parse the recognized voice commands. The model is trained based on large-scale language data and can recognize the operator's intentions and generate appropriate task instructions based on specific scenarios.

[0057] For example, when the operator issues the instruction “start the loader and prepare for excavation”, the NLP model can extract the two operations of “start” and “excavation” and generate corresponding execution instructions.

[0058] Working condition analysis: The system monitors the working status of the loader in real time, including the current operation progress, load condition, fuel consumption status, etc. Based on these data, the system can determine which operations can be performed immediately and which operations need to wait.

[0059] For example, if the loader is performing other operations, the system may delay executing new instructions to avoid task conflicts.

[0060] Task priority adjustment: The system dynamically adjusts the order of task execution based on the importance of the instructions. Tasks with higher priority (such as emergency braking or equipment failure handling) are executed immediately, while other tasks may wait in line.

[0061] The system ensures that the operator's key instructions are completed first by dynamically adjusting tasks, avoiding equipment overload or task conflicts.

[0062] Generate feedback: After parsing the instructions and adjusting the task order, the system will generate corresponding feedback information to inform the operator of the execution status of the current task.

[0063] Feedback can include current execution status, estimated completion time, or warning messages to help operators understand the operating status of the equipment.

[0064] Task Execution: Finally, the parsed and adjusted instructions are passed to the execution module to ensure that the operator's commands can be executed at the most appropriate time. The system executes tasks according to the job status and monitors the progress of tasks in real time.

[0065] 4. Text-to-speech (TTS) module Implementation principle: After the operator issues an instruction, the system needs to provide real-time feedback to the operator to ensure that the operator understands the working status of the loader. The TTS module in the present invention can convert the generated text feedback into speech and adjust the volume, speech speed and tone of the speech output according to the noise level of the loader's surrounding environment to ensure that the feedback information can be heard clearly.

[0066] content: The main function of the text-to-speech module is to convert text feedback into speech and adjust the speech parameters according to the real-time noise environment, so that the operator can clearly hear the feedback in different environments. The system dynamically adjusts the volume and speech speed to ensure that the voice prompts can still be clearly received in a noisy environment.

[0067] step: Noise monitoring: The system can monitor the noise level in the current environment in real time through noise sensors installed in the cab or other noise-sensitive locations. According to the changes in noise, the system will dynamically adjust the voice output parameters.

[0068] Dynamic Adjustment: The system automatically adjusts the volume and speech speed of TTS output according to the real-time noise level. In a high-noise environment, the system increases the voice volume and speeds up the speech to ensure that the voice prompts are not masked by background noise; in a quiet environment, the system lowers the volume to avoid voice interference.

[0069] For example, when the loader is performing high-noise operations (such as loading and moving heavy objects), the system will increase the voice volume and clarity; when the loader is stopped or in a low-noise environment, the system automatically lowers the volume to reduce the operator's auditory burden.

[0070] Generate spoken feedback: The parsed text information (such as device status, task progress, warning information, etc.) is used to generate voice prompts through the TTS engine. The TTS engine uses synthesized sound to convert text content into fluent and natural speech.

[0071] The system can adjust the content of voice feedback according to the complexity of the task. For urgent tasks, the voice prompts will be shorter and clearer, while for ordinary tasks, the feedback can be more detailed.

[0072] Voice playback: Finally, the generated voice feedback is played to the operator through the vehicle audio system, ensuring that the operator can receive equipment status updates or other important information in a timely manner.

[0073] 5. Personalized speech synthesis module Implementation principle: In order to provide a personalized voice interaction experience, this module allows operators to set and select the style of voice output according to their personal preferences. Operators can customize parameters such as speech speed, volume, and tone to ensure that voice prompts that meet personal preferences are obtained in long-term operations. Through this function, the system can provide different operators with a unique voice interaction experience, enhancing the comfort and efficiency of operation.

[0074] content: The core function of the personalized speech synthesis module is to provide personalized options for speech output, allowing operators to set parameters such as speech speed, volume, and intonation according to their personal needs. Operators can adjust these settings through the vehicle system interface, and the system will generate voice prompts that meet their preferences based on the operator's selections.

[0075] step: User Preferences: Operators can customize the parameters of voice output through the system interface, including: Speech rate: Operators can select slow, normal or fast speech rate to suit different operating needs.

[0076] Volume: Operators can adjust the volume of voice output so that they can clearly hear the voice prompts in different working environments.

[0077] Intonation: Operators can choose different voice tones to ensure that voice prompts are more in line with personal habits in different operating scenarios.

[0078] These settings will be recorded by the system and associated with the operator's personal account so that their personalized settings can be restored each time they log in.

[0079] 2. Personalized voice selection: The system provides a variety of preset voice styles for operators to choose from, such as: Concise mode: Provides concise and clear voice feedback to operators who prefer shorter voice prompts, reducing unnecessary prompts.

[0080] Detailed Mode: Provides detailed voice feedback for operators who require more detailed operation guidance, ensuring that every operation step is clearly explained.

[0081] Operators can choose the appropriate voice style according to their personal needs to ensure comfort and efficiency in long-term use. Through the system's user interface, operators can switch voice styles at any time, which is convenient and quick.

[0082] 3. Personalized voice generation: Based on the operator's preferences and personalized selections, the system will generate voice prompts that meet the operator's needs through the text-to-speech (TTS) module. Whether it is a concise prompt mode or a detailed voice guidance, the system will adjust the output according to the operator's selection.

[0083] 4. Automatic Application: The system will automatically apply the previously set personalized voice configuration each time the operator logs in, ensuring consistent voice style and settings each time it is used.

[0084] The intelligent vehicle-mounted voice interaction system of the present invention solves the voice recognition problem, voice interaction experience problem and intelligent control problem of complex operation tasks in loader operation through the integration of multiple modules. Through technologies such as noise processing, voice recognition, natural language processing and personalized voice synthesis, the system can improve the accuracy of voice recognition in complex operating environments, optimize the operator's work experience, and significantly improve the automation and intelligence level of the loader.

[0085] This system is not only suitable for loaders, but can also be widely used in other heavy machinery and equipment, such as excavators, bulldozers, mining equipment, etc. Through the introduction of this system, the intelligent operation level of construction machinery will be greatly improved, and the operator's work efficiency and operation safety will also be significantly improved.

[0086] The examples described in the present invention are merely descriptions of the preferred implementation modes of the present invention, and are not intended to limit the concept and scope of the present invention. Without departing from the design concept of the present invention, various modifications and improvements made to the technical solutions of the present invention by engineers and technicians in this field should all fall within the protection scope of the present invention.

Claims

1. A loader vehicle-mounted voice interaction system, characterized in that: The system includes a noise collection and processing module, a speech signal collection and recognition module, a natural language processing and dialogue management module, a text-to-speech module, and a personalized speech synthesis module, among which: The noise collection and processing module collects noise data in the environment in real time through noise sensors installed at multiple key locations of the loader, and uses Kalman filtering, Wiener filtering and spectrum subtraction algorithms to process noise in different frequency bands, ensuring that the voice signal is separated from the noise and transmitted to the voice recognition module in a clear state; The voice signal acquisition and recognition module uses a deep learning model to recognize the noise-reduced voice signal. The system uses voice activation detection technology to ensure the extraction of valid voice signals and uses ASR voice recognition to convert them into executable text instructions. The natural language processing and dialogue management module analyzes the recognized voice commands based on NLP technology, and dynamically adjusts the task priority based on the current working conditions of the loader, generates commands and executes the corresponding tasks; The text-to-speech module dynamically adjusts the volume, speed and tone of voice prompts according to the real-time noise environment of the loader, ensuring that the operator can clearly receive feedback information in different noise environments.

2. The loader vehicle-mounted voice interaction system according to claim 1, characterized in that: The noise collection and processing module uses the Kalman filter algorithm to process low-frequency noise, the Wiener filter to process intermediate-frequency noise, and the spectrum subtraction method to process high-frequency noise.

3. The loader vehicle-mounted voice interaction system according to claim 1, characterized in that: The voice signal acquisition and recognition module uses voice activation detection technology to distinguish effective voice signals from background noise signals to improve the accuracy and efficiency of voice command recognition.

4. The loader vehicle-mounted voice interaction system according to claim 1, characterized in that: The natural language processing and dialogue management module monitors the loader's operating status, load conditions, fuel consumption and other data in real time, intelligently analyzes the current working conditions and prioritizes the execution of key task instructions.

5. The loader vehicle-mounted voice interaction system according to claim 1, characterized in that: The text-to-speech module dynamically adjusts output parameters according to the real-time noise environment, increasing the volume and speech speed when the loader is in a high-noise state and reducing the volume in a quiet environment to ensure the clarity and adaptability of the voice feedback.

6. The loader vehicle-mounted voice interaction system according to claim 1, characterized in that: It also includes a personalized speech synthesis module that allows the operator to set different voice prompt styles, including concise mode and detailed mode, and automatically saves the operator's preference settings to maintain a consistent operating experience during long-term use.

7. A method for using the loader vehicle-mounted voice interaction system as claimed in claim 1, characterized in that: Includes steps: The noise is collected through sensors and divided into low-frequency noise, medium-frequency noise and high-frequency noise; Adaptive Kalman filter algorithm is used to process the low-frequency noise of the loader voice signal; The intermediate frequency noise of the loader speech signal is processed by Wiener filtering; The high-frequency noise of the loader's voice signal is processed by spectrum subtraction; Extract and recognize the de-noised language signal to generate text instructions; The recognized text instructions are matched with the internal instruction library of the loader. If the matching degree between the recognized text and the known instructions is higher than the threshold, the instructions will be verified. Parse verified instructions using pre-trained NLP models; Get the working status of the loader and determine the priority of the instruction; The task execution order is dynamically adjusted according to the priority of the instructions and passed to the execution module for execution.

8. The method for using the loader vehicle-mounted voice interaction system according to claim 7, characterized in that: The step of dynamically adjusting the task execution order according to the priority of the instructions and passing them to the execution module for execution also includes: after adjusting the task order, generating corresponding feedback information to inform the operator of the execution status of the current task.

9. The method for using the loader vehicle-mounted voice interaction system according to claim 8, characterized in that: Feedback information is converted from text to voice playback, including: Detect the noise level in the current environment; Automatically adjust the output volume and speech speed according to the real-time noise level; The parsed text information generates voice prompts through the TTS engine; The generated voice feedback is played to the operator via the vehicle audio system.

Images