A method for obtaining ideal delta pulse excitation using high-frequency pulse sound source

Abstract

The present invention discloses a method for obtaining an ideal delta pulse excitation using a high-frequency pulse sound source, comprising the steps of: recording a high-frequency electric spark pulse sound source in a fully anechoic chamber and recording the signal as a WAV file; performing inverse filtering on the WAV file and saving the obtained inverse filter as a WAV file; performing a binaural impulse response test using the high-frequency electric spark pulse sound source and a scaled artificial head in a scaled model, and saving the scaled binaural impulse response file as a WAV file; generating n binaural impulse responses that are approximate to those obtained when the ideal delta pulse signal is used as the sound source; performing scale ratio conversion to obtain the binaural impulse response corresponding to the full-scale hall; and performing phase-locked averaging to obtain a binaural impulse response that can be used in subjective listening experiments. The present invention can obtain a binaural impulse response that is approximate to that obtained when the ideal delta pulse signal is used as the sound source, further improve the bandwidth and accuracy, and simultaneously improve the signal-to-noise ratio, making the obtained signal more suitable for auralization experiments.

Description

Technical Field

[0001] The present invention relates to the field of architectural acoustic scale model auralization technology and audio-visual integration technology based on architectural acoustic scale model technology, and in particular to a method for obtaining ideal delta pulse excitation by using a high-frequency electric spark pulse sound source in acoustic scale model testing. Background Art

[0002] Scaled model experiments are one of the main predictive methods in architectural acoustics. Scaled model experiments involve reducing a real hall to a certain scale, using a high-frequency sound source to generate sound, and recording the impulse responses in the model to calculate the hall's acoustic parameters and analyze its acoustic defects. Based on the results of scaled model experiments, architects can adjust the hall's shape and eliminate acoustic defects in the early stages of design. Therefore, scaled model experiments in architectural acoustics are an important auxiliary tool for hall sound quality design. Currently, scaled model experiments are commonly used to measure single-channel impulse responses and predict single-channel acoustic parameters. Using scaled artificial heads of the same scale, binaural impulse responses can be obtained for calculating the hall's spatial perception parameters (IACC) and for auralization experiments. Sun Haitao et al. disclosed in Chinese invention patent CN103895086B, "A method for producing a 1:10 scaled artificial head for architectural acoustics," which details the design and production of a 1:10 scaled artificial head. This method can be used to produce a 1:10 scale artificial head and use it to measure the binaural correlation coefficient (IACC) of the hall spatial perception parameter and record the scaled binaural impulse response.

[0003] Commonly used high-frequency sound sources internationally are spark pulse sound sources and small 12-sided spherical sound sources. High-frequency spark pulse sound sources generate sound by storing charge in a capacitor. Triggering a switch to discharge the capacitor generates a high voltage across the electrodes, quickly ionizing the air between them. The previously insulating air becomes conductive, and the stored charge in the capacitor is rapidly discharged, instantly flowing a large current between the electrodes. The air temperature around the high-voltage electrode rises rapidly. Due to the short discharge time, the energy converted from the high-voltage discharge to heat energy cannot be transferred through the air. Therefore, the air around the electrode expands rapidly, generating a high-intensity sound pulse. High-frequency spark pulse sound sources are characterized by short pulse width, high sound pressure level, good omnidirectionality, and excellent repeatability. The small 12-sided spherical sound source is designed based on the traditional 12-sided spherical sound source used in actual hall steady-state testing. Each small speaker on the sound source emits a steady-state high-frequency ultrasonic signal. Polyvinylidene fluoride (PVDF) film is typically used as the sound-generating material. Because the sound pressure level of a small 12-sided sound source is not high enough, the measurement frequency band is narrow. In 2020, the famous Korean scholar Jin Yong Jeon published a paper in the journal Building and Environment titled "Objective and subjective assessment of sound diffuseness in musical venues via computer simulations and a scale model." The paper used computer simulations and a 1:10 scale model to conduct a subjective and objective evaluation of the scattering of musical spaces. In the scaled model experiment, the author used a small 12-sided spherical sound source as a point sound source (frequency response range 1-50kHz) and obtained a scaled binaural impulse response through a swept frequency measurement method. During the experiment, nitrogen was filled to maintain constant air humidity. The frequency range of the signal used in the final listening experiment was 100Hz to 4000Hz, which is not wide enough for music evaluation.

[0004] There are several main methods for obtaining impulse responses in scaled halls: maximum-length sequence (MLS) pseudo-random noise (PSR) measurement, frequency sweep measurement, and spark pulse measurement. The first two methods are based on digital signal source delta pulses. Sound is emitted from a small dodecahedron source, recorded with a scaled artificial head at the receiving point, and binaural impulse responses are obtained through deconvolution. However, due to the low sound pressure level and frequency response of the small dodecahedron source, the measurement frequency range is limited, failing to meet the bandwidth requirements for auralization. In actual testing, spark pulses can be used to obtain binaural impulse responses with a wider spectrum. The spark pulse method uses a high-frequency spark source to generate sound, recorded with a scaled artificial head at the receiving point, and then the binaural impulse response is obtained. Spark sources offer high sound pressure levels, wide frequency responses, and good repeatability. The resulting binaural impulse responses have a high signal-to-noise ratio and a wide frequency response, making them more suitable for auralization experiments. However, the time-domain spectrum of the spark source differs from the ideal delta pulse signal, resulting in certain limitations. Summary of the Invention

[0005] In order to solve the problems existing in the prior art, the present invention provides a method for obtaining delta pulse excitation by using a high-frequency electric spark pulse sound source in acoustic scale model testing.

[0006] The present invention provides a method for obtaining an ideal delta pulse excitation using a high-frequency pulse sound source, comprising the steps of:

[0007] Step 1: Record the high-frequency spark pulse sound source in a fully anechoic chamber, record the recording signal as a WAV file, and measure the directivity and linear attenuation characteristics of the high-frequency spark pulse sound source;

[0008] Step 2: performing inverse filtering on the Wav file recorded in the anechoic chamber, and saving the obtained inverse filter as a Wav file;

[0009] Step 3. Perform binaural impulse response testing on the scaled model using a high-frequency electric spark pulse sound source and a scaled artificial head. Repeat the test n times at the same measuring point and save the measured scaled binaural impulse response file as a WAV file.

[0010] Step 4. Convolve the n scaled binaural impulse response files at the same measurement point with the inverse filter of the electric spark pulse sound source to generate n binaural impulse responses that are close to those obtained by using the ideal delta pulse signal as the sound source;

[0011] Step 5. Perform scale conversion on the n convolved scaled binaural impulse responses to obtain the binaural impulse responses corresponding to the full-scale hall;

[0012] Step 6. Phase-locked averaging is performed on the aligned direct sounds of the n converted full-scale binaural impulse responses to reduce noise and obtain a binaural impulse response that can be used in subjective listening experiments.

[0013] Furthermore, in step 1, a high-frequency sound analyzer B&K Lanxi module is used to perform recording in a fully anechoic room.

[0014] Furthermore, in step 5, Dirac 5.5 software is used to perform scale ratio conversion.

[0015] In order to make the impulse response test results obtained by the spark sound source close to the excitation response of the delta pulse in the hall obtained by convolving the MLS signal and the swept frequency signal, the spark signal is processed as follows:

[0016] (1) Normalize the spark signal recorded in the anechoic chamber;

[0017] (2) Import the normalized impulse response Wav file into the Aurora plug-in of Audition software and use the Invert Kirkeby function to perform inverse filtering. Set the upper and lower limits of the filter and save the obtained inverse filter as a Wav file.

[0018] (3) The inverse transform of the EDM dry signal is convolved with the impulse response recorded in the scaled model to obtain a binaural impulse response that approximates the delta pulse excitation. However, since the background noise in the hall impulse response recorded in the scaled model is also convolved with the inverse transform during the convolution process, increasing the background noise of the system, it is necessary to measure multiple times in the scaled model and average the impulse responses multiple times to reduce the background noise.

[0019] Furthermore, the theoretical basis is as follows:

[0020]

[0021]

[0022]

[0023]

[0024] Where: x δ (t) is the delta pulse, h(t) is the hall response, and y(t) is the binaural impulse response obtained by measuring the delta pulse; x d (t) is the high-frequency electric spark signal pulse, x n (t) is the inverse filtering of the high-frequency spark signal pulse, y d(t) is the binaural impulse response in the scaled model measured by the spark signal; Formula (1) indicates that the standard binaural impulse response y(t) of the system is the convolution of the δ pulse and the hall response, Formula (2) indicates that the convolution of the high-frequency spark signal pulse and its inverse filter is the δ pulse, Formula (3) indicates that the binaural impulse response obtained by directly measuring the high-frequency spark signal pulse is the convolution of the high-frequency spark signal pulse and the hall response, and Formula (4) indicates that the binaural impulse response y obtained by directly measuring the high-frequency spark signal pulse is the convolution of the high-frequency spark signal pulse and the hall response. d (t) and the inverse filtering of the high-frequency spark signal pulse x n (t) can be convolved to obtain the binaural impulse response y(t) obtained by the delta pulse measurement.

[0025] Furthermore, in step 6, after phase-locked averaging, the signal-to-noise ratios (INR) of the two binaural channels are greater than 35 dB and the SNR is greater than 16 dB.

[0026] Furthermore, the scale ratio of the scaled model is 1:10, the scale ratio of the scaled artificial head for measuring binaural impulse response is 1:10, and the frequency range of the obtained full-scale binaural impulse response is 20 Hz-10000 Hz.

[0027] Compared with the prior art, the present invention can achieve at least the following beneficial effects:

[0028] This paper proposes a method for generating delta pulse excitation using a high-frequency electric spark pulse source in acoustic scale model testing, utilizing digital signal processing techniques. This technique produces binaural impulse responses that approximate those obtained using an ideal delta pulse signal as the source, further improving bandwidth and accuracy. Multiple averaging also increases the signal-to-noise ratio of the binaural pulses, making the resulting signals more suitable for auralization experiments.

[0029] The present invention can be applied to scale model sonification technology and audio-visual integration technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 Flow chart of the steps of this method.

[0031] Figure 2 A diagram of a 1:10 scale model test system provided by an embodiment of the present invention.

[0032] Figure 3 This is a time domain diagram of the high-frequency electric spark pulse sound source measured in an anechoic chamber provided by an embodiment of the present invention.

[0033] Figure 4 In an embodiment of the present invention, a high-frequency electric spark pulse measured in an anechoic chamber is normalized and then imported into the Aurora plug-in of Audition software, and an inverse filtered time domain graph is obtained using the Invert Kirkeby function.

[0034] Figure 5 The delta pulse signal is obtained by convolving the high-frequency electric spark pulse sound source provided by the embodiment of the present invention with its inverse filtering.

[0035] Figure 6 This is a linear attenuation diagram of the spectrum of a high-frequency electric spark pulse sound source measured in an anechoic chamber provided by an embodiment of the present invention.

[0036] Figure 7 This is a schematic diagram of the positions of head measurement points at a 1:10 scale provided by an embodiment of the present invention.

[0037] Figure 8 This is a schematic diagram of the subjective listening experiment scoring provided by an embodiment of the present invention.

[0038] Figure 9 This is a statistical table of subjective evaluation results provided by an embodiment of the present invention, wherein (a) is a statistical diagram of subjective evaluation scores of realism, and (b) is a statistical diagram of subjective evaluation scores of clarity. DETAILED DESCRIPTION

[0039] The purpose of the present invention will be described in further detail below with reference to the accompanying drawings and specific examples. The examples cannot be described one by one here, but the implementation methods of the present invention are not limited to the following examples.

[0040] The present invention provides a method for obtaining delta pulses using a high-frequency electric spark pulse sound source in acoustic scale model testing, comprising the steps of:

[0041] Step 1: Record the high-frequency spark pulse sound source in a fully anechoic chamber and save the recorded signal as a WAV file. To verify the spark sound source's omnidirectional nature and its 6dB attenuation with doubling of distance, the directivity and linear attenuation characteristics of the high-frequency spark pulse sound source were also measured.

[0042] In some embodiments of the present invention, the directivity of the high-frequency electric spark pulse sound source is shown in Table 1 and Table 2, and the linear attenuation characteristics of the point sound source are shown in Table 2. Figure 6 .

[0043] The high-frequency electric spark pulse sound source has good non-directivity, linear attenuation characteristics of a point sound source, and stability of repeated sounding.

[0044] In some embodiments of the present invention, an electric spark pulse sound source emits sound in a fully anechoic chamber, and then a high-frequency sound analyzer B&K Lanxi module is used to record the high-frequency electric spark pulse sound source in the fully anechoic chamber. The analysis frequency of the Lanxi module is 100 kHz, and the signal is received through a microphone, wherein the test microphone uses a B&K 1 / 8-inch microphone 4138.

[0045] Table 1 Average sound pressure level of each measuring point on the horizontal plane of the spark pulse sound source

[0046]

[0047] Table 2 Average sound pressure level at each measuring point on the vertical plane of the spark pulse sound source

[0048]

[0049] Step 2. Import the WAV file recorded in the anechoic chamber into the Aurora plug-in of Audition software and perform inverse filtering using the Invert Kirkeby function. Set the upper and lower frequency limits of the inverse transform based on the frequency range of the model analysis, and save the resulting inverse filter as a WAV file.

[0050] In some embodiments of the present invention, the spark signal recorded in the anechoic chamber is normalized before the Wav file is inverse filtered.

[0051] In some embodiments of the present invention, the Wav file (high-frequency electric spark pulse time domain diagram) recorded in the anechoic chamber is shown in FIG. Figure 3 , is the Wav file saved after inverse filtering (inverse filtering time domain diagram of high-frequency EDM pulse) Figure 4 .

[0052] Step 3. Perform binaural impulse response test in a 1:10 scale model of Tianjin Cultural Center Concert Hall using a high-frequency electric spark pulse sound source and a 1:10 scale artificial head (see Figure 7 ), repeat the test n times at the same measuring point, and save the measured scaled binaural impulse response file as a Wav file;

[0053] In some embodiments of the present invention, the same measuring point position is tested repeatedly 200 times, that is, n=200. It can be understood that in other embodiments, n can take other values.

[0054] Step 4. Because the EDM signal has excellent non-directionality and repeatability, convolve the n scaled binaural impulse response files at the same measurement point with the inverse filter of the EDM pulse sound source to generate n binaural impulse responses that approximate those obtained when the ideal delta pulse signal is used as the sound source;

[0055] Step 5. Perform scale conversion on the n convolved scaled binaural impulse responses to obtain the binaural impulse responses corresponding to the full-scale hall;

[0056] In some embodiments of the present invention, Dirac 5.5 software is used to perform scale ratio conversion.

[0057] Step 6. Phase-locked averaging is performed on the aligned direct sounds of the n converted full-scale binaural impulse responses to reduce noise and obtain a binaural impulse response that can be used in subjective listening experiments.

[0058] Repeated measurements at the same measurement point and phase-locked averaging can reduce system noise and improve the INR (Impulse Response to Noise Ratio) and SNR (Signal to Noise Ratio) of the binaural impulse response to meet the requirements of the listening experiment.

[0059] In some embodiments of the present invention, the noise reduction values of the impulse responses after 200 averaging are shown in Tables 3 and 4. After phase-locked averaging, the signal-to-noise ratio (INR) of the two binaural channels is greater than 35 dB and the SNR is greater than 16 dB, respectively. This results in binaural impulse responses that can be used in subjective listening experiments.

[0060] Table 3 Statistics and trend prediction of INR values at different frequencies under different average times

[0061]

[0062] Table 4 Statistics and trend prediction of SNR values at each frequency under different averaging times

[0063]

[0064] To verify the effectiveness of the method of the present invention, the binaural impulse response obtained by the above method, the binaural impulse response without inverse transformation convolution, and the binaural impulse response measured in a real hall were convolved with the band's dry signal, and then a subjective listening experiment was conducted. The subjective experimental results were statistically analyzed. The experimental process details and statistical analysis results are as follows:

[0065] (1) The audio dry signal used in the listening experiment was Mozart Symphony No. 40. The listening experiment participants were 10 students (5 males and 5 females).

[0066] (2) Selected listening evaluation indicators:

[0067] Plausibility refers to maintaining the original timbre and quality of a sound. In this experiment, the plausibility of the auralized signal at different reception points in a scaled model was evaluated using an audio signal convolved with the impulse response of a real recording of the Tianjin Cultural Center Concert Hall as a benchmark.

[0068] Clarity: Horizontal clarity refers to the degree to which successive notes can be separated and distinguished, meaning one can clearly hear the melody of a rapidly and continuously played passage. Vertical clarity refers to the degree to which notes played simultaneously in each part can be distinguished, meaning the transparency and layering of the music.

[0069] (3) Three listening conditions:

[0070] a) Binaural impulse responses recorded in a real auditorium are convolved with the band's dry signal (as a listening standard);

[0071] b) Directly convolve the 200-time averaged scaled model impulse response with the band dry signal (first segment);

[0072] c) The inverse transformation of the spark pulse signal is convolved with the scaled model impulse response, and the binaural impulse response obtained by multiple averaging is convolved with the band dry signal (second paragraph);

[0073] (4) Statistical analysis of subjective experimental results:

[0074] Depend on Figure 9 As can be seen from the statistical table, using the audio obtained by convolving the live recorded impulse response with the band's dry signal (condition a) as the evaluation standard, the subjective evaluation score of the audio obtained in condition b is much higher than that in condition c in terms of both realism and clarity, indicating that the signal obtained by the method of the present invention is more suitable for auralization experiments.

[0075] The above embodiments of the present invention are merely examples for the purpose of clearly illustrating the present invention and are not intended to limit the embodiments of the present invention. Those skilled in the art will appreciate that other variations or modifications may be made based on the above description. It is not necessary and impossible to enumerate all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the claims of the present invention.

Claims

1. A method for obtaining an ideal delta pulse excitation using a high-frequency pulse sound source, characterized in that: Including steps: Step 1: Record the high-frequency spark pulse sound source in a fully anechoic chamber, record the recording signal as a WAV file, and measure the directivity and linear attenuation characteristics of the high-frequency spark pulse sound source; Step 2: performing inverse filtering on the Wav file recorded in the anechoic chamber, and saving the obtained inverse filter as a Wav file; Step 3. Perform binaural impulse response testing on the scaled model using a high-frequency electric spark pulse sound source and a scaled artificial head. Repeat the test n times at the same measuring point and save the measured scaled binaural impulse response file as a WAV file. Step 4. Convolve the n scaled binaural impulse response files at the same measurement point with the inverse filter of the electric spark pulse sound source to generate n binaural impulse responses that are close to those obtained by using the ideal delta pulse signal as the sound source; Step 5. Perform scale conversion on the n convolved scaled binaural impulse responses to obtain the binaural impulse responses corresponding to the full-scale hall; Step 6. Phase-locked averaging is performed on the aligned direct sounds of the n converted full-scale binaural impulse responses to reduce noise and obtain a binaural impulse response that can be used in subjective listening experiments.

2. The method for obtaining an ideal delta pulse excitation using a high-frequency pulse sound source according to claim 1, characterized in that: In step 1, a high-frequency sound analyzer B&K Lanxi module was used to record the sound in a fully anechoic room.

3. The method for obtaining an ideal delta pulse excitation using a high-frequency pulse sound source according to claim 1, characterized in that: In step 2, the Invert Kirkeby function is used for inverse filtering.

4. The method for obtaining an ideal delta pulse excitation using a high-frequency pulse sound source according to claim 1, characterized in that: In step 2, the spark signal recorded in the anechoic chamber is first normalized when performing inverse filtering.

5. The method for obtaining an ideal delta pulse excitation using a high-frequency pulse sound source according to claim 4, characterized in that: In step 2, the Wav file is imported into the Aurora plug-in of Audition software and inverse filtered using the Invert Kirkeby function.

6. The method for obtaining an ideal delta pulse excitation using a high-frequency pulse sound source according to claim 1, characterized in that: In step 5, Dirac 5.5 software is used to perform scale ratio conversion.

7. The method for obtaining an ideal delta pulse excitation using a high-frequency pulse sound source according to claim 1, characterized in that: The frequency range of the obtained full-scale binaural impulse response is 20Hz-10000Hz.

8. The method for obtaining an ideal delta pulse excitation using a high-frequency pulse sound source according to claim 1, characterized in that: In step 3, the scale ratio of the scaled model is 1:10, and the scale ratio of the scaled artificial head for measuring binaural impulse response is 1:

10.

9. The method for obtaining an ideal delta pulse excitation using a high-frequency pulse sound source according to claim 1, characterized in that: In step 6, after phase-locked averaging, the signal-to-noise ratios (INR) of the two binaural channels are greater than 35 dB and the SNR is greater than 16 dB.

10. A method for obtaining an ideal delta pulse excitation using a high-frequency pulse sound source according to any one of claims 1 to 9, characterized in that: The theoretical basis is Where: x δ (t) is the delta pulse, h(t) is the hall response, and y(t) is the binaural impulse response obtained by measuring the delta pulse; x d (t) is the high-frequency electric spark signal pulse, x n (t) is the inverse filtering of the high-frequency spark signal pulse, y d (t) is the binaural impulse response in the scaled model measured using the spark signal; t is time; Formula (1) indicates that the standard binaural impulse response y(t) of the system is the convolution of the δ pulse and the hall response. Formula (2) indicates that the convolution of the high-frequency EDM signal pulse and its inverse filter is the δ pulse. Formula (3) indicates that the binaural impulse response obtained by directly measuring the high-frequency EDM signal pulse is the convolution of the high-frequency EDM signal pulse and the hall response. Formula (4) indicates that the binaural impulse response y obtained by directly measuring the high-frequency EDM signal pulse is the convolution of the high-frequency EDM signal pulse and the hall response. d (t) and the inverse filtering of the high-frequency spark signal pulse x n (t) can be convolved to obtain the binaural impulse response y(t) obtained by the delta pulse measurement.

Images