A speech signal enhancement method and system

Abstract

The application relates to the technical field of voice processing, in particular to a voice signal enhancement method and system, which comprises the following steps: obtaining real-time voice signals and historical voice signals of a target area; filtering the real-time voice signals of the target area by using an improved least mean square algorithm to enhance voice quality; introducing the concept of an adaptive initial step, dynamically adjusting the step according to the maximum eigenvalue of the autocorrelation matrix of the real-time signals and the step adjustment degree to adapt to the change of the voice signals; the step adjustment degree reflects the convergence trend of the voice signal filtering process; the difference degree is used for quantifying the difference between the real-time voice signals and the historical voice signals, which can effectively evaluate the characteristics of the voice signals and dynamically adjust the filtering process. The application solves the problem that the existing algorithm is difficult to realize ideal convergence efficiency and filtering effect under different scenes.

Description

Technical Field

[0001] This invention relates to the field of speech processing technology. More specifically, this invention relates to a speech signal enhancement method and system. Background Technology

[0002] With the widespread application of voice communication and speech recognition technologies, the quality of voice signals is becoming increasingly important in various systems. However, in real-world environments, voice signals are often affected by noise, echoes, interference, and other factors, leading to a decline in voice quality and potentially impacting system operation and user experience. Especially in low signal-to-noise ratio environments, noise mixes with the voice signal, and traditional voice enhancement methods face problems such as insufficient denoising performance and voice distortion. Therefore, effectively extracting clear voice information from noisy voice signals has become a critical issue that urgently needs to be addressed in the field of voice processing.

[0003] Currently, speech signal enhancement methods mainly include time-domain and frequency-domain methods. Time-domain methods primarily process the signal waveform, but they have limitations in noise suppression. Frequency-domain methods, by converting the signal into the spectral domain for analysis, can improve speech quality to some extent. However, in complex environments, the performance of these methods is still constrained by various factors. Therefore, there is an urgent need for a new speech signal enhancement method that, through more precise signal processing, improves speech clarity and intelligibility, thereby enhancing the performance of speech communication and speech recognition systems.

[0004] The least mean square (LMS) algorithm, as a core technology of adaptive filtering, performs exceptionally well in speech signal processing. Its dynamic adaptability, low computational complexity, and hardware friendliness effectively improve speech quality in various scenarios. However, in practical applications, especially in conference room recording scenarios, the speech signal captured by the microphone is often subject to multipath interference such as wall reflections, resulting in background noise. When using the LMS algorithm for speech enhancement, the choice of step size is crucial. An excessively large step size may increase steady-state error, thus affecting the filtering effect; while an excessively small step size will cause the algorithm to react slowly in dynamic environments, failing to adapt to changes in the speech signal in a timely manner, leading to slower convergence. This makes it difficult for existing algorithms to achieve ideal convergence efficiency and filtering effects in different scenarios. Summary of the Invention

[0005] To address the problem that existing algorithms mentioned in the background art are unable to achieve ideal convergence efficiency and filtering effect in different scenarios, the present invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a speech signal enhancement method, comprising: acquiring a real-time speech signal and a historical speech signal of a target region; filtering the real-time speech signal using an improved least mean square algorithm to enhance the real-time speech signal, and obtaining the enhanced real-time speech signal; wherein the improved least mean square algorithm includes an adaptive initial step size, the adaptive initial step size being positively correlated with the degree of difference between the real-time speech signal and the historical speech signal, the step size adjustment degree, and the initial step size; the step size adjustment degree characterizes the filtering convergence trend of the speech signal.

[0007] The above technical solution improves the step size of the least mean square algorithm by dynamically adjusting the step size according to the degree of difference and convergence trend of the speech signal, so that the filtering process can maintain stability and efficiency in different speech environments. This solves the problem that existing algorithms are difficult to achieve ideal convergence efficiency and filtering effect in different scenarios.

[0008] Furthermore, the adaptive initial step size for: In the formula, To indicate the degree of difference, For step size adjustment degree, The largest eigenvalue of the autocorrelation matrix of the real-time speech signal. This is the initial step size.

[0009] Furthermore, the degree of difference for, In the formula, The first in real-time voice signal Each signal amplitude, It is the average value of the amplitudes of all signals in the historical speech signal. The standard deviation of the amplitude of all signals in the historical speech signal is given. This represents the total number of signal amplitudes in the real-time speech signal. Z These are the zero-crossing counts of the historical voice signal and the real-time voice signal, respectively. This represents the total number of historical speech signals.

[0010] Furthermore, the step size adjustment degree , In the formula, , These are the first in the historical speech signal. The error signal value, the first One error signal value, This represents the maximum value of the error signal in the real-time speech signal. This represents the total number of error signal values ​​in the historical speech signal. These are the preset hyperparameters.

[0011] The above technical solution introduces a step size adjustment factor to adaptively optimize the step size of the least mean square algorithm, enabling the filtering process to more accurately adapt to the dynamic changes in the speech signal. The calculation method of the step size adjustment factor comprehensively considers the distribution of error signals in historical speech signals, weights the accumulated errors using a normalized approach, and incorporates a smoothing factor to ensure that the step size adjustment effectively reflects the changing trend of signal errors. This method, while maintaining filtering stability, improves the algorithm's adaptability to noisy environments, allowing the step size to maintain sufficient adjustment space when the signal changes drastically, while avoiding over-adjustment when the signal tends to stabilize, thereby improving the convergence speed and robustness of the filtering.

[0012] Furthermore, the step size adjustment degree , In the formula, , These are the first in the historical speech signal. The error signal value, the first One error signal value, This represents the maximum value of the error signal in the real-time speech signal. This represents the total number of error signal values ​​in the historical speech signal. These are the preset hyperparameters.

[0013] The above technical solution improves filtering performance by optimizing the calculation method of step size adjustment, enabling the least mean square algorithm to adapt more accurately to error changes in different speech environments. It utilizes the distribution characteristics of error signal values ​​in historical speech signals, introduces normalization processing, combines weighted adjustment of error accumulation, and uses a logarithmic function to smooth the step size changes, ensuring that the step size adjustment reflects error dynamics while avoiding instability caused by excessive numerical fluctuations.

[0014] Furthermore, the initial step size , , It represents the largest eigenvalue of the autocorrelation matrix of the real-time speech signal.

[0015] The above technical solution sets the initial step size to the reciprocal of the largest eigenvalue of the autocorrelation matrix of the real-time speech signal, enabling the least mean square algorithm to adaptively adjust the step size under different signal environments, thus ensuring the stability and efficiency of the filtering process. By utilizing the eigenvalue characteristics of the autocorrelation matrix, the step size can adapt to the statistical characteristics of the signal, achieving dynamic adjustment and avoiding instability caused by an excessively large step size or a decrease in convergence speed caused by an excessively small step size.

[0016] Furthermore, the acquisition of real-time and historical voice signals from the conference room specifically involves: collecting real-time and historical voice data by deploying microphones in the conference room, and performing equal sampling and discretization processing on the collected real-time and historical voice data to obtain real-time and historical voice signals.

[0017] Furthermore, the error signal value is specifically defined as follows: the predicted speech signal obtained from the historical speech signal is obtained using a long short-term memory network, and the difference between the historical speech signal and the predicted speech signal is used as the error signal value.

[0018] Furthermore, it also includes performing spectral subtraction processing on the enhanced real-time voice signal.

[0019] The above technical solution enhances the real-time voice signal and then introduces spectral subtraction processing. Spectral subtraction further optimizes the signal-to-noise ratio of the signal, making the voice features more prominent. Especially in complex or low signal-to-noise ratio environments, it can significantly reduce the interference caused by background noise, enabling applications such as voice recognition and voice interaction to obtain more accurate input data.

[0020] In a second aspect, the present invention provides a speech signal enhancement system, including a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a speech signal enhancement method as described above is implemented.

[0021] The beneficial effects of this invention are as follows:

[0022] This invention, by combining an improved least mean square algorithm, dynamically optimizes the filtering process of speech signals, improving their clarity and quality. Through a combination of adaptive initial step size, step size adjustment, and difference level, this invention can perform precise filtering based on the characteristics of real-time and historical speech signals, ensuring rapid convergence and effective noise suppression under varying environmental noise conditions. This not only improves the quality of the speech signal but also enhances the adaptability and robustness of the speech signal enhancement system, making it widely applicable in fields such as speech recognition, voice communication, and intelligent voice interaction. Attached Figure Description

[0023] Figure 1 This is a flowchart schematically illustrating a speech signal enhancement method according to an embodiment of the present invention;

[0024] Figure 2 This is a schematic block diagram illustrating the structure of a speech signal enhancement system according to an embodiment of the present invention. Detailed Implementation

[0025] An embodiment of a speech signal enhancement method.

[0026] like Figure 1 The flowchart shown below illustrates a speech signal enhancement method according to an embodiment of the present invention, which includes the following steps:

[0027] S1: Acquire real-time and historical voice signals of the target area.

[0028] In one embodiment, real-time and historical voice signals within the target area are acquired to support subsequent functions such as voice analysis, meeting record generation, or voice recognition. Specifically, taking a conference room as an example, the steps include:

[0029] First, multiple high-sensitivity microphone arrays are deployed in the conference room to cover the entire effective pickup range, ensuring comprehensive acquisition of speech data from different locations and speakers during the meeting. The microphone arrays can capture real-time speech data as well as historical speech data stored over a period of time, forming a complete speech signal database.

[0030] Secondly, the acquired speech data is subjected to equal sampling and discretization to obtain a time-stable speech signal. The sampling frequency can be set to 10Hz by default to ensure sufficient time resolution with low computational cost, thus preserving the key features of the speech signal. Of course, the sampling frequency can be dynamically adjusted according to actual application needs. For example, the sampling frequency can be appropriately increased in high-precision speech analysis scenarios, while it can be decreased when resources are limited or time resolution requirements are not high, in order to reduce data storage and computational overhead.

[0031] To improve the quality of the speech signal and reduce environmental noise interference, a high-pass filter is first applied to the acquired real-time speech signal. The high-pass filter effectively removes low-frequency noise, such as background noise from air conditioners, projection equipment, and power lines, thereby improving the clarity and intelligibility of the speech signal. Furthermore, high-pass filtering reduces the impact of low-frequency resonance on speech features, enabling subsequent algorithms such as speech recognition, speaker separation, or speech enhancement to more accurately extract speech content and speaker features.

[0032] S2: The real-time speech signal is filtered using an improved least mean square algorithm to enhance the real-time speech signal.

[0033] In one embodiment, the improved least mean square algorithm includes an adaptive initial step size, the adaptive initial step size... for: In the formula, To indicate the degree of difference, For step size adjustment degree, The maximum eigenvalue is the autocorrelation matrix of the real-time speech signal. The calculation of the maximum eigenvalue of the real-time speech signal autocorrelation matrix is ​​obtained using existing technology and will not be elaborated upon in this solution. This is the initial step size;

[0034] By introducing an improved least mean square algorithm with an adaptive initial step size, the step size can be intelligently adjusted according to the dynamic changes of the speech signal, thereby improving the convergence speed and stability of the filtering. Taking into account the degree of difference in the speech signal, the step size adjustment trend, and the signal autocorrelation characteristics, the step size can be quickly adjusted to an appropriate range when there are large changes, accelerating the adaptability of the filtering process. Conversely, when the changes are small, the step size range is limited to prevent instability caused by over-adjustment. Through this adaptive optimization, the algorithm can maintain efficient noise suppression capabilities under different noise environments while preserving the key features of the speech signal to the greatest extent, making the enhanced speech signal clearer and more accurate, and providing high-quality input data for applications such as speech recognition and intelligent speech processing.

[0035] It should be noted that the initial step size is adjusted based on the degree of difference and the degree of step size adjustment. The greater the degree of difference, the greater the gap between the latest speech signal and the historical speech signal; while the greater the degree of step size adjustment, the worse the filtering effect of the initial step size, and the more step size is needed to speed up the steady-state convergence speed of the filter.

[0036] The degree of difference for, In the formula, The first in real-time voice signal Each signal amplitude, It is the average value of the amplitudes of all signals in the historical speech signal. The standard deviation of the amplitude of all signals in the historical speech signal is given. This represents the total number of signal amplitudes in the real-time speech signal. Z These are the zero-crossing counts of the historical voice signal and the real-time voice signal, respectively. This represents the total number of historical speech signals.

[0037] By introducing a statistical feature-based method for calculating the degree of difference, the filtering process can more accurately reflect the changing trends between real-time and historical speech signals. It comprehensively considers the dispersion of signal amplitude, overall mean deviation, and changes in the number of zero-crossings to measure the fluctuation characteristics of the real-time speech signal compared to historical speech signals. Through this calculation method, the algorithm can increase the step size adjustment amplitude when signal changes are large, accelerating filter convergence, while appropriately reducing the convergence step size when the signal is stable, avoiding signal distortion caused by over-correction.

[0038] The step size adjustment , In the formula, , These are the first in the historical speech signal. The error signal value, the first One error signal value, This represents the maximum value of the error signal in the real-time speech signal. This represents the total number of error signal values ​​in the historical speech signal. These are the preset hyperparameters.

[0039] By optimizing the step size adjustment method, the filtering algorithm can more accurately adapt to the error distribution characteristics of historical speech signals, achieving dynamic step size adjustment to improve convergence efficiency and robustness. Taking into account the cumulative impact of historical error signal values ​​and utilizing normalization, the step size adjustment can flexibly reflect error change trends while avoiding excessive influence of extreme errors on the adjustment process. By introducing a smoothing factor, the step size adjustment is ensured to respond to error fluctuations while maintaining overall stability, thus achieving optimal filtering results under various noise environments. This optimization strategy enhances the algorithm's noise resistance and adaptability, making speech signal enhancement more accurate and effectively reducing noise interference.

[0040] The initial step size , , The initial step size is the largest eigenvalue of the autocorrelation matrix of the real-time speech signal. By correlating the initial step size with the largest eigenvalue of the autocorrelation matrix, the startup process of the filtering algorithm is optimized, allowing the step size to more accurately match the characteristics of the signal in the early stages of signal processing. This setting enables the step size to be automatically adjusted according to the autocorrelation of the speech signal, ensuring that the filtering algorithm can quickly adapt to the dynamic changes of the signal in the initial stage, avoiding instability or slow convergence problems caused by excessively large or small step sizes.

[0041] The error signal value is specifically defined as follows: the predicted speech signal obtained from the historical speech signal using a long short-term memory network, and the difference between the historical speech signal and the predicted speech signal is used as the error signal value.

[0042] In another embodiment, the step size adjustment degree , In the formula, , These are the first in the historical speech signal. The error signal value, the first One error signal value, This represents the maximum value of the error signal in the real-time speech signal. This represents the total number of error signal values ​​in the historical speech signal. These are the preset hyperparameters.

[0043] By optimizing the step size adjustment and combining it with the error signal characteristics and dynamic changes of historical speech signals, the step size in the filtering process can be adjusted more precisely to adapt to signal fluctuations and noise characteristics. This method considers the cumulative effect of error values ​​and makes the step size adjustment smoother and more adaptive through weighted calculation of historical errors, thereby improving the stability and convergence efficiency of the filtering algorithm. Introducing logarithmic smoothing of error values ​​further enhances the adaptability to different types of signals, avoids over-adjustment caused by excessively large step sizes, and ensures that appropriate adjustments are maintained even with small errors.

[0044] S3: Obtain the enhanced real-time voice signal.

[0045] In one embodiment, after the enhanced real-time speech signal undergoes initial enhancement processing, it is further subjected to spectral subtraction to further improve the quality of the speech signal and reduce noise interference. Spectral subtraction, a common noise suppression technique, effectively suppresses the influence of background noise by removing estimated noise components from the speech signal's spectrum. This processing step can significantly improve the clarity and audibility of the speech signal, especially in noisy environments. Specifically, in the enhanced real-time speech signal, by analyzing its spectral characteristics, estimated values ​​of noise components are extracted and subtracted from the signal's spectrum, thereby achieving effective noise suppression.

[0046] Spectral subtraction processing can further eliminate residual noise in speech signals, especially in high-noise environments such as conference rooms, streets, or industrial areas, improving the purity of the speech signal. This not only aids in speech signal recognition and processing but also improves the accuracy and response speed of systems in applications such as intelligent speech recognition and voice interaction. The application of spectral subtraction processing can also, to some extent, avoid speech distortion or recognition errors caused by excessive noise, ensuring that the enhanced speech signal is clearer and more natural, thus improving the user experience.

[0047] This invention, by combining an improved least mean square algorithm, effectively improves the quality of real-time audio signals in conference rooms. First, filtering enhances the real-time audio signal, and an adaptive step-size adjustment method ensures stable and efficient convergence of the filtering process, thereby optimizing the clarity and audibility of the audio signal. Furthermore, the calculation of the degree of difference adaptively adjusts the step size based on the dynamic characteristics of the audio signal, avoiding the limitations of static parameters and making the algorithm more adaptable and flexible. In addition, spectral subtraction further reduces background noise interference, ensuring the purity of the audio signal, especially in noisy environments, enhancing the accuracy and stability of speech recognition. Overall, this invention significantly improves the quality of audio signals and is suitable for various real-time audio processing scenarios such as conferences and speech recognition, maintaining high-quality audio output, especially in complex noisy environments.

[0048] An embodiment of a speech signal enhancement system:

[0049] like Figure 2 As shown, a block diagram of a speech signal enhancement system according to an embodiment of the present invention includes a processor and a memory.

[0050] This invention also provides a speech signal enhancement system. For example... Figure 2 As shown, the system includes a processor and a memory, the memory storing computer program instructions, which, when executed by the processor, implement the speech signal enhancement method according to the present invention.

[0051] The aforementioned voice signal enhancement system also includes other components well known to those skilled in the art, such as communication interfaces. Their settings and functions are known in the art and will not be described in detail here.

[0052] In this invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc., or any other medium that can be used to store desired information and can be accessed by an application, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device. Any application or module described in this invention can be implemented using computer-readable / executable instructions stored or otherwise maintained by such a computer-readable medium.

[0053] In the description of this specification, "multiple" or "several" means at least two, such as two, three or more, unless otherwise explicitly specified.

[0054] While this specification has shown and described numerous embodiments of the invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of this invention.

Claims

1. A speech signal enhancement method characterized by, The method comprises the following steps: acquiring real-time voice signals and historical voice signals of a target area; filtering the real-time voice signals by using an improved least mean square algorithm to enhance the real-time voice signals, and obtaining enhanced real-time voice signals; wherein the improved least mean square algorithm comprises an adaptive initial step size, and the adaptive initial step size is positively correlated with a difference degree between the real-time voice signals and the historical voice signals, a step size adjustment degree and an initial step size; the step size adjustment degree represents a filtering convergence trend of the voice signals.

2. The speech signal enhancement method of claim 1, wherein, the adaptive initial step size is: , wherein, is the degree of difference, is the step size adjustment degree, is the maximum eigenvalue of the real-time speech signal autocorrelation matrix, is the initial step size.

3. The speech signal enhancement method of claim 1, wherein, The degree of difference for, In the formula, The first in real-time voice signal Each signal amplitude, It is the average value of the amplitudes of all signals in the historical speech signal. The standard deviation of the amplitude of all signals in the historical speech signal is given. This represents the total number of signal amplitudes in the real-time speech signal. Z These are the zero-crossing counts of the historical voice signal and the real-time voice signal, respectively. This represents the total number of historical speech signals.

4. The speech signal enhancement method of claim 1, wherein, The step size adjustment , In the formula, , These are the first in the historical speech signal. The error signal value, the first One error signal value, This represents the maximum value of the error signal in the real-time speech signal. This represents the total number of error signal values ​​in the historical speech signal. These are the preset hyperparameters.

5. The speech signal enhancement method of claim 1, wherein, The step size adjustment , In the formula, , These are the first in the historical speech signal. The error signal value, the first One error signal value, This represents the maximum value of the error signal in the real-time speech signal. This represents the total number of error signal values ​​in the historical speech signal. These are the preset hyperparameters.

6. The speech signal enhancement method of claim 1, wherein, the initial step size , , is the maximum eigenvalue of the autocorrelation matrix of the real-time speech signal.

7. The speech signal enhancement method of claim 1, wherein, The acquiring of the real-time voice signals and the historical voice signals of the target area specifically comprises: collecting real-time voice data and historical voice data by arranging microphones in a conference room, and performing equal sampling and discretization processing on the collected real-time voice data and historical voice data to obtain the real-time voice signals and the historical voice signals.

8. A speech signal enhancement method according to claim 4 or 5, characterized in that, The error signal value specifically comprises: obtaining a predicted voice signal of the historical voice signal by using a long short-term memory network, and taking a difference between the historical voice signal and the predicted voice signal as the error signal value.

9. The speech signal enhancement method of claim 1, wherein, The method further comprises performing frequency spectrum subtraction processing on the enhanced real-time voice signals.

10. A speech signal enhancement system characterized by The method comprises a memory and a processor, and the memory stores computer program instructions, which, when executed by the processor, implement the voice signal enhancement method of any one of claims 1-9.

Images