A method, system and Bluetooth speaker for intelligent control of a Bluetooth speaker

Abstract

This invention relates to the field of speaker device technology, and more particularly to a method, system, and Bluetooth speaker for intelligent control of a Bluetooth speaker. The method includes: determining whether the audio latency meets requirements; in response to a playback state not meeting requirements, storing the duration and transmission parameters in a fault dataset; training a probability model; predicting the probability and duration of a playback state not meeting requirements in the next cycle; in response to a probability greater than a preset probability, adjusting the buffer depth based on the duration; in response to a probability less than or equal to a preset probability, obtaining the proportion of human voices in the audio file; and adjusting the gain or buffer depth of the mid-to-high frequency components in the audio based on the proportion of human voices. By utilizing the strong correlation between historical transmission parameters and latency faults to train a multi-task probability model, the risk of latency exceeding limits and the duration in the next cycle are predicted, enabling graded and precise control based on the prediction results, thereby increasing the operational stability of the Bluetooth speaker.

Description

Technical Field

[0001] This invention relates to the field of speaker equipment technology, and in particular to a method, system and Bluetooth speaker for intelligent control of Bluetooth speakers. Background Technology

[0002] With the rapid development of voice recognition technology, Bluetooth speakers have become increasingly popular due to their small size, low power consumption, and portability. However, Bluetooth speakers cannot directly connect to the internet; they need to establish a Bluetooth connection with a mobile phone or other device first, sending the audio data to be played to the speaker via Bluetooth.

[0003] Patent publication number CN110418245A discloses a method, apparatus, terminal device, and computer-readable storage medium for reducing the response latency of a Bluetooth speaker. The method includes: upon receiving a wake-up message, performing a wake-up interrupt operation to wake up the Bluetooth speaker; after the Bluetooth speaker is woken up, caching the received voice data; detecting whether invalid data exists in the cached voice data; if invalid data exists in the cached voice data, determining that the data other than the invalid data in the cached voice data is valid data. This technical solution suffers from data caching issues when the Bluetooth signal fluctuates, which in turn affects the audio latency of the audio file, impacting the user experience and resulting in low stability. Summary of the Invention

[0004] The purpose of this invention is to provide a method, system, and Bluetooth speaker for intelligent control of Bluetooth speakers, in order to solve the problem in the prior art that when Bluetooth signals fluctuate, data caching is affected, which in turn affects the audio latency of audio files, impacting the user experience and resulting in low stability.

[0005] This invention provides a method for intelligent control of a Bluetooth speaker, comprising:

[0006] Real-time acquisition of audio latency from Bluetooth speakers;

[0007] The playback status is determined based on the audio delay to determine whether it meets the requirements.

[0008] In response to the audio delay being greater than the delay threshold, it is determined that the playback status does not meet the requirements, and the duration of the playback status not meeting the requirements and the transmission parameters when the playback status does not meet the requirements are stored in the fault dataset.

[0009] Train a probability model based on the fault dataset;

[0010] Based on the transmission parameters in the current cycle and the probability model, the probability of the playback state not meeting the requirements in the next cycle and the duration of the playback state not meeting the requirements are predicted. In response to the probability being greater than a preset probability, the buffer depth is adjusted according to the duration. In response to the probability being less than or equal to the preset probability, the proportion of human voices in the audio file is obtained, and the speaker operating parameters are adjusted according to the proportion of human voices.

[0011] As a preferred technical solution for a Bluetooth speaker intelligent control method, the step of responding to the audio delay being greater than a delay threshold, determining that the playback state does not meet the requirements, and storing the duration of the playback state not meeting the requirements and the transmission parameters when the playback state does not meet the requirements into a fault dataset, including:

[0012] The fault level is determined based on the extent to which the audio delay exceeds the threshold;

[0013] Differentiated storage weights are assigned to different levels of faults; the greater the audio latency exceeds the threshold, the higher the storage weight.

[0014] Data is stored in descending order of weight value, and high-weight data is retained first when storage space is insufficient.

[0015] As a preferred technical solution for intelligent control of Bluetooth speakers, the probability model is established based on the duration of the playback state not meeting the requirements and the corresponding transmission parameters as the training set. The probability model is trained using a multi-task learning method. If the transmission parameters are input, the probability of the playback state not meeting the requirements and the duration of the playback state not meeting the requirements are output.

[0016] As a preferred technical solution for a Bluetooth speaker intelligent control method, the method predicts the probability of the playback state not meeting the requirements in the next cycle and the duration of the playback state not meeting the requirements based on the transmission parameters in the current cycle and the probability model. In response to the probability being zero, the audio sampling rate is increased to a higher level.

[0017] In response to the probability being less than or equal to the preset probability and not being zero, the human voice and background noise in the audio file are separated, and the proportion of human voice in the total audio is calculated and recorded as the human voice proportion. The operating parameters to be adjusted are determined based on the human voice proportion, including the gain or buffer depth of the mid-to-high frequency part of the audio.

[0018] In response to the probability being greater than the preset probability, the cache depth is adjusted according to the duration.

[0019] As a preferred technical solution for intelligent control of Bluetooth speakers, the step of determining the adjustable operating parameters based on the proportion of human voice includes:

[0020] In response to the human voice proportion exceeding the proportion threshold, a frequency band equalizer is used to perform gain processing on the mid-to-high frequency part of the audio signal;

[0021] If the proportion of human voices is less than or equal to the aforementioned proportion threshold, the cache depth is reduced.

[0022] As a preferred technical solution for intelligent control of a Bluetooth speaker, the step of adjusting the buffer depth according to the duration includes:

[0023] Obtain the duration of the prediction;

[0024] Select several time windows with a duration of duration from the historical playback records;

[0025] Calculate the amount of data for the audio file played within each time window, and determine the maximum value of the data.

[0026] Based on the maximum value of the data volume and combined with a smooth transition algorithm, the current cache depth is adjusted to the target cache depth, and subsequent audio data is preloaded;

[0027] When caching the maximum value of the data volume, the target cache depth is 1.2 times or more of the required cache depth.

[0028] As a preferred technical solution for intelligent control of Bluetooth speakers, the transmission parameters are transmission parameters within a preset time period before the audio delay exceeds the delay threshold, including: signal strength, retransmission rate, and throughput fluctuation rate.

[0029] As a preferred technical solution for intelligent control of Bluetooth speakers, the real-time acquisition of the audio latency of the Bluetooth speaker includes:

[0030] Embed timestamps into audio data packets at the sending end of the audio file;

[0031] Obtain the local system time at which the audio data packet begins decoding and processing, calculate the time difference between the time corresponding to the timestamp and the local system time at which the audio data packet begins decoding and processing, and obtain the initial delay;

[0032] The initial delay is processed by Kalman filtering to eliminate the influence of measurement noise and obtain the final audio delay.

[0033] The present invention also provides a Bluetooth speaker intelligent control system, comprising:

[0034] A detection module is used to obtain the audio latency of the Bluetooth speaker;

[0035] The analysis module is used to determine whether the playback status meets the requirements based on the audio delay;

[0036] The data collection module, in response to the audio delay being greater than the delay threshold, determines that the playback status does not meet the requirements, and stores the duration of the playback status not meeting the requirements and the transmission parameters when the playback status does not meet the requirements into the fault dataset.

[0037] The training module is used to train a probabilistic model based on the fault dataset;

[0038] The adjustment module is used to predict the probability of the playback state not meeting the requirements in the next period and the duration of the playback state not meeting the requirements based on the transmission parameters in the current period and the probability model. In response to the probability being greater than a preset probability, the buffer depth is adjusted according to the duration. In response to the probability being less than or equal to the preset probability, the proportion of human voices in the audio file is obtained, and the speaker operating parameters are adjusted according to the proportion of human voices.

[0039] The present invention also provides a Bluetooth speaker, including a digital signal processor, a main control processor, a memory, and a computer program stored in the memory and executable on the main control processor. When the computer is executed by the processor, it implements the steps of the above-described intelligent control method for the Bluetooth speaker.

[0040] Compared with the prior art, the beneficial effect of the present invention is that by training a probability prediction model with the operating parameters when the playback state does not meet the requirements stored in the fault data, the probability and duration of the future playback state not meeting the requirements can be predicted based on the current operating parameters. The dual analysis and decision-making combined with the proportion of human voice achieves a balance between audio latency and sound quality optimization, thereby increasing the stability of the Bluetooth speaker during operation.

[0041] Furthermore, when the probability is greater than a preset probability, the maximum data size of the audio file played within a continuous duration is determined through historical playback records. The cache depth is then adjusted based on this maximum data size, dynamically expanding the cache depth and preloading data. This avoids audio latency by playing preloaded data before it becomes unacceptable, proactively building an "anti-fluctuation buffer." When the probability is less than or equal to a preset probability and not zero, the proportion of human voices in the audio is separated, and the mid-to-high frequency gain is enhanced (for voice scenarios) or the cache depth is reduced (for music scenarios), achieving an optimal balance between sound quality and latency. In applications, this can also be combined with the speaker's underlying hardware for collaborative optimization: using timestamp filtering to eliminate monitoring errors and using a dedicated DSP to accelerate human voice separation. This prevents audio playback interruptions caused by network fluctuations or content switching when Bluetooth data transmission fluctuates, further increasing the stability of the Bluetooth speaker during operation. Attached Figure Description

[0042] Figure 1This is a flowchart illustrating the steps of the Bluetooth speaker intelligent control method according to an embodiment of the present invention;

[0043] Figure 2 This is a structural block diagram of the Bluetooth speaker intelligent control system according to an embodiment of the present invention. Detailed Implementation

[0044] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0045] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0046] Please see Figure 1 The diagram shown is a flowchart of the steps of the Bluetooth speaker intelligent control method according to an embodiment of the present invention, including:

[0047] Step S1: Obtain the audio latency of the Bluetooth speaker in real time;

[0048] Step S2: Determine whether the playback status of the Bluetooth speaker meets the requirements based on the audio delay;

[0049] Step S3: In response to the audio delay being greater than the delay threshold, it is determined that the playback status does not meet the requirements, and the duration of the playback status not meeting the requirements and the transmission parameters when the playback status does not meet the requirements are stored in the fault dataset.

[0050] Step S4: Train a probability model based on the fault dataset;

[0051] Step S5: Based on the transmission parameters and probability model in the current period, predict the probability of the playback status not meeting the requirements in the next period and the duration of the playback status not meeting the requirements. If the probability is greater than the preset probability, adjust the buffer depth according to the duration. If the probability is less than or equal to the preset probability, obtain the proportion of human voice in the audio file and adjust the speaker operating parameters according to the proportion of human voice.

[0052] In detail, a probability prediction model is trained by storing the operating parameters of the fault dataset when the playback state does not meet the requirements. This enables the prediction of the probability and duration of future playback state failures based on the current operating parameters. Combined with the proportion of human voices, a dual analysis and decision-making process is used to achieve a balance between audio latency and sound quality optimization, thereby increasing the stability of the Bluetooth speaker during operation.

[0053] Furthermore, determining whether the audio delay meets the requirements includes:

[0054] If the audio delay is less than or equal to the delay threshold, the requirement is met.

[0055] If the audio delay exceeds the delay threshold, it is determined that the playback status does not meet the requirements. The duration of the playback status not meeting the requirements and the transmission parameters when the playback status does not meet the requirements are obtained and stored in the fault dataset.

[0056] In implementation, the latency threshold is determined based on actual usage needs. The maximum acceptable latency range for the human ear varies in different scenarios. For example, for everyday music playback, the human ear is less sensitive to latency. If there is no visual synchronization requirement (e.g., no video feed), a latency of 200-300ms generally will not significantly affect the auditory experience. For video viewing and calls, the human ear is more sensitive to audio-visual asynchrony, and the maximum acceptable latency is usually 150-200ms. If the latency exceeds 200ms, audio-visual asynchrony will be clearly noticeable, leading to a degraded experience. In gaming scenarios, such as action games or real-time interactive scenarios, players have extremely high requirements for the latency of operation feedback, usually needing to control the latency below 100ms, ideally below 50ms, to avoid a disconnect between operation and sound feedback. Preferably, the latency threshold in this embodiment of the invention is 150ms.

[0057] Further, in step S3, if the playback status does not meet the requirements, the duration of the playback status failure and the transmission parameters when the playback status fails are stored in the fault dataset, including:

[0058] Step S31: Classify the fault level according to the magnitude of the audio delay exceeding the threshold;

[0059] Step S32: Assign differentiated storage weights to different levels of faults. The greater the audio delay exceeds the threshold, the higher the storage weight.

[0060] Step S33: Store data in descending order of weight value. When storage space is insufficient, prioritize retaining data with high weight.

[0061] In implementation, the fault levels are divided into three levels: Level I: audio latency exceeds the threshold by less than or equal to 50ms; Level II: audio latency exceeds the threshold by more than 50ms but less than or equal to 100ms; Level III: audio latency exceeds the threshold by more than 100ms. Differentiated storage weights are assigned to different fault levels, where: Level I: weight 1; Level II: weight 3; Level III: weight 5.

[0062] In detail, by employing a fault level classification and differentiated weight storage strategy for the data in the fault dataset, the system can prioritize the retention of high-risk fault data, effectively utilize limited storage space, and ensure that key fault features are not discarded. This improves the accuracy of model training, increases the accuracy of predicting the probability of playback status not meeting requirements in the next cycle, and the duration of such unacceptable playback status. Based on the prediction results, the speaker's operating parameters are adjusted to increase the speaker's ability to resist playback status not meeting requirements, thereby increasing the stability of the Bluetooth speaker during operation.

[0063] Further, in step S4, the probabilistic model is established using the duration of the playback state not meeting requirements and the corresponding transmission parameters as the training set. The probabilistic model is trained using a multi-task learning method. If the transmission parameters are input, the model outputs the probability of the playback state not meeting requirements and the duration of the playback state not meeting requirements. In practical applications, there is a direct influence between transmission parameters and audio latency. Therefore, it is feasible to obtain the mapping relationship between the probability of the playback state not meeting requirements, the duration of the playback state not meeting requirements, and the transmission parameters by collecting transmission parameter data and audio latency data for training. The process of training and establishing the model using known data is existing technology and will not be elaborated here.

[0064] Specifically, by jointly learning the feature representation of shared transmission parameters through multi-tasks, the accuracy of probability classification and duration regression is optimized simultaneously. The model can simultaneously output the fault risk level and duration of the next cycle, providing a quantitative basis for cache depth adjustment. This avoids excessive or insufficient cache adjustment that could lead to audio-visual desynchronization or resource waste, maintaining stable and efficient playback, thereby further increasing the stability of the Bluetooth speaker during operation.

[0065] Further, in step S5, based on the transmission parameters in the current period and the probability model, the probability of the playback status not meeting the requirements in the next period and the duration of the playback status not meeting the requirements are predicted, including:

[0066] In response to a probability of zero, the audio sampling rate is increased to a higher level;

[0067] Specifically, in practical applications, the commonly used audio sampling rate is:

[0068] 8kHz is commonly used for low-quality voice communication.

[0069] 16kHz is commonly used in speech coding and speech recognition to improve speech clarity.

[0070] 32kHz is commonly used for broadcasting, low-bandwidth audio of medium quality, and is suitable for voice.

[0071] 44.1kHz, commonly used in high-fidelity audio formats such as CD, MP3, and WAV, is suitable for music.

[0072] 48kHz is a standard commonly used for video and audio synchronization in film and video production.

[0073] 96kHz is commonly used in high-resolution audio applications where high-quality audio is desired.

[0074] 192kHz is commonly used in professional recording and high-fidelity audio with ultra-high resolution, pursuing the ultimate sound quality.

[0075] Preferably, the audio sampling rates, from low to high, are: 8kHz, 16kHz, 32kHz, 44.1kHz, 48kHz, 96kHz, and 192kHz.

[0076] In response to a probability less than or equal to a preset probability and not zero, the human voice and background noise in the audio file are separated, and the proportion of human voice in the total audio is calculated and recorded as the human voice proportion. The operating parameters to be adjusted are determined based on the human voice proportion, including the gain or buffer depth of the mid-to-high frequency part of the audio.

[0077] If the probability is greater than the preset probability, adjust the cache depth according to the duration.

[0078] Further, in step S5, the operating parameters for adjustment are determined based on the proportion of human voices, including:

[0079] In response to the human voice proportion exceeding the proportion threshold, a frequency band equalizer is used to perform gain processing on the mid-to-high frequency part of the audio signal;

[0080] In implementation, the calculation of the proportion of human voice is achieved through audio signal processing technology. This calculation is implemented using a deep neural network sound source separation model, which runs locally on the playback end, separating speech and background sound components in real time. The proportion threshold is a preset value based on factors such as user experience, playback environment, and device performance. The proportion threshold can be adjusted according to real-time audio data. Preferably, in this embodiment, the proportion threshold is set to 40%. The gain of the mid-to-high frequency components in the audio is adjusted using an audio equalizer: by adjusting the audio equalizer, gain adjustment is performed within the 1kHz to 5kHz frequency band. This frequency band is the main frequency region for most languages ​​and human voices. Therefore, by increasing the gain in this frequency band, the performance of human voices can be improved, making them clearer and easier to identify. The specific range of gain (3-6dB) is dynamically adjusted based on the actual proportion of human voices. The adjustment of the frequency band equalizer is based on the characteristics of the audio file, the acoustic performance of the current device, and the real-time network conditions. If the proportion of human voices reaches the threshold, the mid-to-high frequency gain is gradually increased to optimize the sound effect. The adjustment range of the frequency gain is dynamically adjusted based on the current network bandwidth, Bluetooth signal quality, and noise interference in the playback environment. The audio equalizer and the adjustment process of the audio equalizer are existing technologies and will not be described in detail here.

[0081] When the proportion of human voices is less than or equal to the proportion threshold, the cache depth is reduced.

[0082] Specifically, when the proportion of human voices is low, the system shortens the playback time of the audio stream by reducing the buffer depth to optimize audio playback latency. This operation can reduce playback latency and improve the real-time performance of the audio stream, avoiding audio stuttering or delay. The buffer depth only needs to be reduced to the minimum value that does not affect the playback status. Dynamically adjusting the buffer depth is an existing technology in the sound system adjustment process and will not be elaborated on here.

[0083] Furthermore, the cache depth is increased based on the duration, including:

[0084] Obtain the duration of the prediction;

[0085] Select several time windows with a duration of duration from the historical playback records;

[0086] Calculate the amount of data for the audio file played within each time window, and determine the maximum value of the data.

[0087] Based on the maximum data volume and combined with a smooth transition algorithm, the current cache depth is adjusted to the target cache depth, and subsequent audio data is preloaded;

[0088] When caching the maximum value of the data volume, the target cache depth is 1.2 times or more of the required cache depth. The historical playback record refers to the playback history of audio files after this Bluetooth connection.

[0089] In detail, the target cache depth is set at 1.2 times the maximum historical playback data volume, and combined with a smooth transition algorithm, the cache adjustment process is made imperceptible to the user. Actual testing shows that the algorithm can complete the cache depth adjustment within 200ms, with audio quality fluctuations of less than 0.5dB during the adjustment process. This dynamic caching mechanism effectively addresses sudden data fluctuations in Bluetooth transmission, ensuring smooth playback while avoiding resource waste or excessive latency issues caused by traditional fixed caching strategies, thereby further increasing the stability of the Bluetooth speaker during operation.

[0090] Specifically, the transmission parameters are the transmission parameters within a preset time period before the audio delay exceeds the delay threshold, including: signal strength, retransmission rate, and throughput volatility.

[0091] Furthermore, real-time acquisition of the Bluetooth speaker's audio latency includes:

[0092] Synchronize the local clock between the Bluetooth speaker and the data transmitter;

[0093] Embed timestamps into audio data packets at the sending end of the audio file;

[0094] Obtain the local system time when the audio data packet started decoding and processing, calculate the time difference between the time corresponding to the timestamp and the local system time when the audio data packet started decoding and processing, and obtain the initial delay;

[0095] The initial delay is processed by Kalman filtering to eliminate the influence of measurement noise and obtain the final audio delay.

[0096] In detail, this invention employs Kalman filtering to process audio delay data, effectively eliminating measurement noise caused by Bluetooth signal fluctuations. Actual testing shows that this method can reduce delay measurement error from ±20ms to within ±5ms, improving the system's sensitivity to minute delay changes. Accurate delay measurement is fundamental to intelligent control; the application of Kalman filtering enables the system to detect delay change trends earlier, allowing for more timely adjustment decisions and further increasing the stability of the Bluetooth speaker during operation.

[0097] Please see Figure 2 As shown, it is a structural block diagram of the Bluetooth speaker intelligent control system according to an embodiment of the present invention, including:

[0098] A detection module is used to obtain the audio latency of the Bluetooth speaker;

[0099] The analysis module is used to determine whether the playback status meets the requirements based on the audio delay;

[0100] The data collection module, in response to the audio delay being greater than the delay threshold, determines that the playback status does not meet the requirements, and stores the duration of the playback status not meeting the requirements and the transmission parameters when the playback status does not meet the requirements into the fault dataset.

[0101] The training module is used to train a probabilistic model based on the fault dataset;

[0102] The adjustment module is used to predict the probability of the playback state not meeting the requirements in the next period and the duration of the playback state not meeting the requirements based on the transmission parameters in the current period and the probability model. In response to the probability being greater than a preset probability, the buffer depth is adjusted according to the duration. In response to the probability being less than or equal to the preset probability, the proportion of human voices in the audio file is obtained, and the speaker operating parameters are adjusted according to the proportion of human voices.

[0103] Furthermore, embodiments of the present invention also provide a Bluetooth speaker, including a digital signal processor, a main control processor, a memory, and a computer program stored in the memory and executable on the main control processor, wherein the computer, when executed by the processor, implements any of the steps in the above methods.

[0104] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A method for intelligent control of a Bluetooth speaker, characterized in that, include: Real-time acquisition of audio latency from Bluetooth speakers; The playback status is determined based on the audio delay to determine whether it meets the requirements. In response to the audio delay being greater than the delay threshold, it is determined that the playback status does not meet the requirements, and the duration of the playback status not meeting the requirements and the transmission parameters when the playback status does not meet the requirements are stored in the fault dataset. Train a probability model based on the fault dataset; Based on the transmission parameters in the current cycle and the probability model, the probability of the playback state not meeting the requirements in the next cycle and the duration of the playback state not meeting the requirements are predicted. In response to the probability being greater than a preset probability, the buffer depth is adjusted according to the duration. In response to the probability being less than or equal to the preset probability, the proportion of human voices in the audio file is obtained, and the speaker operating parameters are adjusted according to the proportion of human voices.

2. The Bluetooth speaker intelligent control method according to claim 1, characterized in that, In response to the audio delay exceeding a delay threshold, the playback status is determined to be non-compliant. The duration of the non-compliant playback status and the transmission parameters at the time of non-compliant playback status are stored in the fault dataset, including: The fault level is determined based on the extent to which the audio delay exceeds the threshold; Differentiated storage weights are assigned to different levels of faults; the greater the audio latency exceeds the threshold, the higher the storage weight. Data is stored in descending order of weight value, and high-weight data is retained first when storage space is insufficient.

3. The Bluetooth speaker intelligent control method according to claim 1, characterized in that, The probability model is established using the duration of playback status not meeting requirements and the corresponding transmission parameters as the training set. The probability model is trained using a multi-task learning method. If the transmission parameters are input, the model outputs the probability of playback status not meeting requirements and the duration of playback status not meeting requirements.

4. The Bluetooth speaker intelligent control method according to claim 1, characterized in that, The system predicts the probability and duration of the playback status not meeting the requirements in the next period based on the transmission parameters in the current period and the probability model. In response to the probability being zero, the audio sampling rate is increased to a higher level. In response to the probability being less than or equal to the preset probability and not zero, the human voice and background noise in the audio file are separated, and the proportion of human voice in the total audio is calculated and recorded as the human voice proportion. The operating parameters to be adjusted are determined based on the human voice proportion, including the gain or buffer depth of the mid-to-high frequency part of the audio. In response to the probability being greater than the preset probability, the cache depth is adjusted according to the duration.

5. The Bluetooth speaker intelligent control method according to claim 4, characterized in that, The operating parameters determined based on the proportion of human voices include: In response to the human voice proportion exceeding the proportion threshold, a frequency band equalizer is used to perform gain processing on the mid-to-high frequency part of the audio signal; If the proportion of human voices is less than or equal to the aforementioned proportion threshold, the cache depth is reduced.

6. The Bluetooth speaker intelligent control method according to claim 4, characterized in that, The step of adjusting the cache depth according to the duration includes: Obtain the duration of the prediction; Select several time windows with a duration of duration from the historical playback records; Calculate the amount of data for the audio file played within each time window, and determine the maximum value of the data. Based on the maximum value of the data volume and combined with a smooth transition algorithm, the current cache depth is adjusted to the target cache depth, and subsequent audio data is preloaded; When caching the maximum value of the data volume, the target cache depth is 1.2 times or more of the required cache depth.

7. The Bluetooth speaker intelligent control method according to claim 2, characterized in that, The transmission parameters are the transmission parameters within a preset time period before the audio delay exceeds the delay threshold, including: signal strength, retransmission rate, and throughput volatility.

8. The Bluetooth speaker intelligent control method according to claim 1, characterized in that, The real-time acquisition of the Bluetooth speaker's audio latency includes: Embed timestamps into audio data packets at the sending end of the audio file; Obtain the local system time at which the audio data packet begins decoding and processing, calculate the time difference between the time corresponding to the timestamp and the local system time at which the audio data packet begins decoding and processing, and obtain the initial delay; The initial delay is processed by Kalman filtering to eliminate the influence of measurement noise and obtain the final audio delay.

9. A Bluetooth speaker intelligent control system, used to implement the Bluetooth speaker intelligent control method according to any one of claims 1-8, characterized in that, include: A detection module is used to obtain the audio latency of the Bluetooth speaker; The analysis module is used to determine whether the playback status meets the requirements based on the audio delay; The data collection module, in response to the audio delay being greater than the delay threshold, determines that the playback status does not meet the requirements, and stores the duration of the playback status not meeting the requirements and the transmission parameters when the playback status does not meet the requirements into the fault dataset. The training module is used to train a probabilistic model based on the fault dataset; The adjustment module is used to predict the probability of the playback state not meeting the requirements in the next period and the duration of the playback state not meeting the requirements based on the transmission parameters in the current period and the probability model. In response to the probability being greater than a preset probability, the buffer depth is adjusted according to the duration. In response to the probability being less than or equal to the preset probability, the proportion of human voices in the audio file is obtained, and the speaker operating parameters are adjusted according to the proportion of human voices.

10. A Bluetooth speaker, characterized in that, It includes a digital signal processor, a main control processor, a memory, and a computer program stored in the memory and executable on the main control processor, wherein the computer, when executed by the processor, implements the steps of the method as described in any one of claims 1 to 8.

Images