Speech airflow pressure signal analysis system

Abstract

The invention discloses a speech airflow pressure signal analysis system. The speech airflow pressure signal analysis system comprises an editing module, a labeling module, a calculation module, a user interaction module and a comparison module. Through a modular system and an automatic processing flow, the analysis efficiency and precision of the speech airflow pressure signal are remarkably improved. The system supports batch analysis of NSP format files and synchronous display of multichannel signals, and the problem of low efficiency of traditional single file processing is solved; an automatic peak-valley value detection and deduplication mechanism is adopted, so that the marking error is effectively reduced, and the data reliability is improved; superposition comparison and linkage analysis of voice, airflow and air pressure signals are realized, and a speech physiological mechanism is visually revealed; the parameter calculation module automatically extracts key indexes such as voice duration, airflow and airflow velocity, and outputs structured data for subsequent statistical application. The system has good compatibility and operability and can be widely applied to the fields of voice research, clinical correction, teaching and the like.

Description

Technical Field

[0001] This invention relates to the field of speech analysis, and more specifically to a speech airflow pressure signal analysis system. Background Technology

[0002] Speech airflow and pressure signals are key physiological signals generated during human articulation. They reflect the dynamic changes in airflow speed, airflow volume, and air pressure during articulation and are the core basis for revealing the mechanism of speech production, conducting phonetics research, and correcting speech disorders.

[0003] In the field of phonetics research, foreign countries began to study the relationship between the glottis and airflow in the 1930s; in China, Professor Fonte of the Royal Institute of Technology in Sweden introduced the theory of speech aerodynamics in the 1980s, which promoted the development of related research in China. Among them, Wu Zongji's (1987) "Experimental Study on the Difference between Aspirated and Inspired Consonants in Mandarin Chinese" revealed the physiological mechanism of Mandarin Chinese consonants for the first time through airflow and air pressure experiments, which became the foundational research in this field in China.

[0004] Although existing technologies have achieved airflow and pressure signal acquisition, significant deficiencies exist in the signal analysis stage, failing to meet research and application needs. Specifically: Insufficient NSP file processing capabilities; the NSP file reading software accompanying the PAS 6600 only supports single-file viewing and cannot process data in batches; furthermore, it cannot specifically extract core parameters such as airflow peaks and pressure troughs, resulting in extremely low operational efficiency when analyzing large samples (e.g., dialect surveys, clinical multi-case statistics). Large signal labeling errors; existing software lacks automatic and accurate labeling functions for airflow and pressure signal peaks and troughs, requiring manual labeling, which is susceptible to subjective influence, leading to errors such as label position misalignment and omissions, thus affecting the accuracy of subsequent parameter calculations. Lack of multi-channel signal collaborative analysis; speech, airflow, and pressure are related physiological indicators of speech production, but existing software does not achieve synchronous display, overlay analysis, and linked labeling of the three channels, making it impossible to intuitively observe the correspondence among the three (e.g., the matching of speech vowels with airflow peaks), limiting in-depth research on speech production mechanisms. Data storage and subsequent analysis are inconvenient. Existing software outputs data in a single format and cannot directly save extracted parameters such as airflow velocity and total airflow into a table format (such as Excel) that is easy to use for statistics. Manual secondary entry is required, which increases data processing costs and is prone to errors.

[0005] In summary, existing technologies suffer from problems such as low processing efficiency, large labeling errors, poor coordination, and difficulty in data reuse. They cannot meet the needs of speech physiology research, clinical correction, and speech teaching for accurate and efficient airflow and air pressure signal analysis. There is an urgent need for a system and method that can achieve multi-channel synchronous processing, automatic and accurate labeling, and parameterized output. Summary of the Invention

[0006] To address the technical problems mentioned above, this invention utilizes a modular system design and a step-by-step analysis method to achieve NSP format file parsing, multi-channel signal processing, automatic annotation, parameter calculation, and visualization analysis.

[0007] To achieve the above objectives, the present invention provides a speech airflow and barometric pressure signal analysis system, comprising: an editing module, an annotation module, a calculation module, a user interaction module, and a comparison module;

[0008] The editing module is used to parse the NSP format file output by the speech airflow and air pressure acquisition device, extract the speech data, airflow data and air pressure data, and convert the data into a multi-channel WAV format file and a MAT file containing metadata.

[0009] The annotation module is used to detect and mark peak points and valley points in selected signal segments of airflow data and air pressure data;

[0010] The calculation module is used to calculate preset physiological parameters based on the marked peak and valley points and signal sampling rate, and save the parameters as a structured data file;

[0011] The user interaction module is built using a graphical user interface to synchronously display voice, airflow, and air pressure signals, and to provide human-computer interaction functions.

[0012] The comparison module is used to generate a superimposed comparison diagram of any two or all of the voice, airflow, and air pressure signals.

[0013] Preferably, the editing module is compatible with NSP format files of various sampling rates, and realizes data parsing and playback by dynamically setting sampling rate variables.

[0014] Preferably, the labeling module marks peak points using the maximum value detection method and valley points using the minimum value detection method, and has a deduplication mechanism. When the error between the position of a newly labeled peak or valley point and the already labeled position is within a preset threshold, the new labeling is skipped.

[0015] Preferably, the physiological parameters calculated by the calculation module include speech duration, airflow rate, and airflow velocity; wherein, the formula for calculating airflow rate is:

[0016] Airflow rate = ∫selectedsignal segment airflow signal value dt,

[0017] Where dt is the sampling time interval;

[0018] The formula for calculating airflow velocity is:

[0019] Airflow velocity = airflow rate / speech duration.

[0020] Preferably, the display area of ​​the user interaction module includes: a voice signal channel, a speech-image channel, an airflow signal channel, and a barometric pressure signal channel arranged from top to bottom, so as to realize the synchronous display and comparison of multi-channel signals.

[0021] Preferably, the superimposed comparison diagram generated by the comparison module synchronously displays the voice signal waveform, airflow signal waveform, and air pressure signal waveform in different colors or line types within the same coordinate axis.

[0022] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0023] This invention significantly improves the efficiency and accuracy of speech airflow and pressure signal analysis through a modular system and automated processing flow. The system supports batch parsing of NSP format files and simultaneous display of multi-channel signals, overcoming the inefficiency of traditional single-file processing. It employs automatic peak-valley detection and deduplication mechanisms to effectively reduce labeling errors and improve data reliability. It enables the superposition, comparison, and linkage analysis of speech, airflow, and pressure signals, intuitively revealing the physiological mechanisms of speech. The parameter calculation module automatically extracts key indicators such as speech duration, airflow rate, and airflow velocity, and outputs structured data for easy subsequent statistical applications. The system possesses excellent compatibility and operability, and can be widely applied in speech research, clinical correction, and teaching, comprehensively solving the problems of low processing efficiency, poor collaboration, and difficulty in data reuse inherent in existing technologies. Attached Figure Description

[0024] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a schematic diagram of the system structure according to an embodiment of the present invention;

[0026] Figure 2 This is a signal labeling display diagram according to an embodiment of the present invention;

[0027] Figure 3 This is a comparison diagram of signal superposition in an embodiment of the present invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0030] Example

[0031] The embodiments of this invention aim to elaborate on the specific implementation of the speech airflow and air pressure signal analysis system and method, so as to help those skilled in the art fully understand the technical solution and application of this invention. The embodiments will be illustrated using the analysis of the Mandarin Chinese "p" sound as an example, combined with the functional description of the system modules and the signal processing flow. However, the scope of application of this invention is not limited to this; any physiological research, clinical evaluation, or teaching application based on speech airflow and air pressure signals can be adjusted and implemented with reference to these embodiments.

[0032] The speech airflow and barometric pressure signal analysis system adopts a modular architecture, including an editing module, an annotation module, a calculation module, a user interaction module, and a comparison module. The overall system workflow begins with the input and parsing of signal data and ends with parameter output and visualization analysis. The modules work collaboratively through data flow to achieve efficient and accurate signal processing. The system design is based on the MATLAB environment, but those skilled in the art can use other programming languages ​​or platforms to implement the same functionality without departing from the core ideas of this invention.

[0033] After system startup, the editing module processes the raw signal files output by the speech airflow and barometric pressure acquisition equipment. Taking an NSP format file acquired by a KAY PAS 6600 airflow and barometric pressure meter as an example, the system parses the file content through the editing module. The NSP file, as a closed format, stores three channels of data: speech, airflow, and pressure. The editing module uses a data reading algorithm to skip invalid information in the file header, extract valid data segments from specified locations, and divide them into three parts: speech data, airflow data, and pressure data. After data parsing, the system converts it into a multi-channel WAV format file and generates a MAT file containing metadata, including sampling rate, signal duration, and channel identifiers. This process supports batch file processing, significantly improving data import efficiency and overcoming the limitations of traditional single-file operations. Furthermore, the editing module has a dynamic sampling rate adaptation function; by adjusting the parsing parameters through system variables, it can be compatible with NSP files output by different acquisition devices, such as signals with sampling rates of 16000Hz or 44100Hz.

[0034] After analysis, the data enters the user interaction module. This module is built using a graphical user interface, providing intuitive signal display and interactive functions. The interface layout includes multiple display channels, arranged from top to bottom as voice signal channel, speech graph channel, airflow signal channel, and air pressure signal channel. The voice signal is normalized, with its amplitude range set to [-1, 1]. The airflow and air pressure signals retain their original physical units (such as mL / s and cmH2O) to facilitate observation of actual signal changes. Users can select specific signal segments for analysis using interface tools; for example, in the "p" tone example, the airflow segment (sampling points 1000 to 3000) can be selected as the target area. The system displays the signal waveform in real time and supports dynamic adjustment of the display range, enabling focusing and zooming of signal segments without manual intervention.

[0035] The annotation module automatically detects and marks the peak and trough points of airflow and air pressure signals based on the user-selected signal segments. The signal annotations are displayed as follows: Figure 2 As shown in the diagram, this module uses a maximum value detection method to identify peak values ​​and a minimum value detection method to identify valley values. During the detection process, the system traverses the data points within the selected segment, locates the extreme value positions, and records their values ​​and sampling point indices. To improve marking accuracy, the module has a built-in deduplication mechanism: when the error between the newly detected peak / valley position and the existing marking is within one sampling point range, the system automatically ignores duplicate markings to avoid data redundancy. For example, in the "p" sound airflow segment, the airflow peak is marked as 280 mL / s (sampling point 1800), and the valley value is marked as 15 mL / s (sampling point 2500); the air pressure peak is marked as 8 cmH2O (sampling point 1750), and the valley value is marked as 1.2 cmH2O (sampling point 2600). The marking results are stored in a MAT file and marked in the display channel with visual elements (such as colored dots) for easy user verification.

[0036] The calculation module automatically calculates multiple physiological parameters based on the annotation results and signal sampling rate. These parameters include speech duration, airflow rate, airflow velocity, maximum airflow value, minimum airflow value, maximum air pressure value, and minimum air pressure value. Speech duration is calculated as the ratio of the number of sampling points in the selected segment to the sampling rate; for example, the duration of the air delivery segment is (3000-1000+1) / 22050 ≈ 0.091 seconds. Airflow rate is obtained by numerically integrating the airflow signal within the selected segment; the integration formula is:

[0037] Airflow rate = ∫ airflow signal value dt

[0038] Where dt is the sampling time interval (1 / sampling rate).

[0039] The airflow velocity is calculated as the quotient of airflow rate and speech duration, i.e.:

[0040] Airflow velocity = airflow rate / speech duration.

[0041] All parameters are saved as double-precision floating-point numbers. The system supports multiple output formats, such as Excel or CSV files. The file name is associated with the original signal file, for example, "parameter-p sound.xls". The output file contains a structured data table with columns including filename, voice duration, maximum airflow value, minimum airflow value, maximum air pressure value, minimum air pressure value, airflow rate, and airflow velocity, facilitating direct import into statistical software for subsequent analysis.

[0042] The comparison module generates a superimposed comparison diagram of multi-channel signals, visually presenting the temporal correlation of speech, airflow, and air pressure signals, such as... Figure 3 As shown, the horizontal axis represents the sampling point sequence. For example, in the aspirated segment of the "p" sound, the overlay clearly shows that the airflow and air pressure reach their peak synchronously within the 1700-2200 sampling point range, confirming the physiological characteristics of the explosive sound. Furthermore, the comparison module can generate a comparison graph of the speech signal and energy curve, calculating speech energy through pre-emphasis filtering and a sliding window to assist users in identifying active speech segments. These visualizations not only enhance the intuitiveness of signal analysis but also provide a reliable basis for research on speech production mechanisms.

[0043] The system is designed with adaptability and scalability in mind. In addition to its core functions, the annotation module supports alternative detection algorithms, such as the sliding window thresholding method, suitable for high-noise signal environments. The calculation module's parameter output format is configurable, compatible with various statistical analysis needs. The user interaction module's interface layout and display options can be customized according to application scenarios, such as highlighting airflow changes in clinical correction or emphasizing air pressure patterns in speech instruction.

[0044] This invention achieves highly efficient analysis of speech airflow and pressure signals through modular integration and automated processing. The system's processing efficiency is more than 50% higher than traditional software, and its labeling error is reduced by 30%. It also supports multi-channel collaborative analysis, providing a powerful tool for speech physiology research, speech disorder assessment, and educational applications. Examples demonstrate that the system successfully extracted key physiological parameters in the analysis of the "p" sound and revealed the intrinsic correlations between signals through visualization, verifying its technical feasibility and practical value.

[0045] The above embodiments are merely examples of specific implementation methods. Those skilled in the art can adjust and extend the module implementation, parameter calculation method, and output format without departing from the core idea of ​​this invention. Any improvements or substitutions based on the spirit of this invention should be covered within the protection scope of this invention.

[0046] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A speech airflow and air pressure signal analysis system, characterized in that, include: Editing module, annotation module, calculation module, user interaction module, and comparison module; The editing module is used to parse the NSP format file output by the speech airflow and air pressure acquisition device, extract the speech data, airflow data and air pressure data, and convert the data into a multi-channel WAV format file and a MAT file containing metadata. The annotation module is used to detect and mark peak points and valley points in selected signal segments of airflow and air pressure data; The calculation module is used to calculate preset physiological parameters based on the marked peak and valley points and signal sampling rate, and save the parameters as a structured data file; The user interaction module is built using a graphical user interface to synchronously display voice, airflow, and air pressure signals, and to provide human-computer interaction functions. The comparison module is used to generate a superimposed comparison diagram of any two or all of the voice, airflow, and air pressure signals.

2. The speech airflow and air pressure signal analysis system according to claim 1, characterized in that, The editing module is compatible with NSP format files of various sampling rates and realizes data parsing and playback by dynamically setting sampling rate variables.

3. The speech airflow and air pressure signal analysis system according to claim 1, characterized in that, The labeling module marks peak points using the maximum value detection method and valley points using the minimum value detection method. It also has a deduplication mechanism. When the error between the position of a newly labeled peak or valley point and the already labeled position is within a preset threshold, the new labeling is skipped.

4. The speech airflow and air pressure signal analysis system according to claim 1, characterized in that, The physiological parameters calculated by the calculation module include speech duration, airflow rate, and airflow velocity; wherein, the formula for calculating airflow rate is: Airflow rate = ∫selectedsignal segment airflow signal value dt, Where dt is the sampling time interval; The formula for calculating airflow velocity is: Airflow velocity = airflow rate / speech duration.

5. The speech airflow and air pressure signal analysis system according to claim 1, characterized in that, The display area of ​​the user interaction module includes, from top to bottom, a voice signal channel, a speech-image channel, an airflow signal channel, and a barometric pressure signal channel, enabling synchronous display and comparison of multiple channel signals.

6. The speech airflow and air pressure signal analysis system according to claim 1, characterized in that, The superimposed comparison diagram generated by the comparison module synchronously displays the voice signal waveform, airflow signal waveform, and air pressure signal waveform in different colors or line types within the same coordinate axis.

Images