A method for synthesizing active automobile sound

Abstract

A method for active automotive sound synthesis employs the following specific steps: Step 1: Extracting sound particles and creating corresponding sound particles based on the vehicle engine operating conditions to form a sound particle library; Step 2: Using a periodic interpolation algorithm to synthesize the sound data to be played; Step 3: Playing the sound data without delay. The automotive sound synthesis method described in this invention utilizes a periodic interpolation particle synthesis algorithm, resolving the "stepped sound" issue in previous particle algorithms. This improves listening comfort, reduces the difficulty of acquiring the original sound of the target vehicle model in the particle algorithm, and enhances the accuracy of sound synthesis. It ensures the coherence and smoothness of the car sound under all vehicle conditions, as well as the continuity of all order lines during software analysis.

Description

Technical Field

[0001] The invention belongs to the technical field of acoustics, and in particular relates to a method for synthesizing active automobile sound. Background Art

[0002] Improving vehicle quality has gradually replaced production capacity growth as the mainstream of the automotive industry. Improving overall acoustic performance has become particularly important, placing increasingly stringent demands on vehicle NVH. With the application of active sound generation technology, sound synthesis algorithms can be used to design in-car sounds that meet target requirements based on user needs. Electronic controllers can then play sounds that change with driving conditions.

[0003] Existing order synthesis methods extract the order characteristics of the target car's sound individually, then increase or decrease them as needed to synthesize the sound that matches the music scheme. While this method can extract some of the target sound's orders, it loses the original car's ambient noise, lacking authenticity. Furthermore, extracting all orders is difficult.

[0004] The granular synthesis method pre-records the target car's sound, then slices it into very short sound particles based on the main order frequency. Finally, the corresponding sound particles are obtained based on the car's actual operating conditions, spliced ​​together, and played back. This method effectively preserves the original sound of the target car, but it places high demands on the recorded audio file. If the audio file is recorded too short, the number of audio particles will be small, reducing the accuracy of the corresponding vehicle's operating status. This can cause stepped sounds in some operating conditions, reducing listening comfort. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention proposes an active car sound synthesis method that addresses the difficulties in extracting original sounds from current particle synthesis methods and the problem of stepped sounds in the synthesized sounds at low speeds. This method ensures that the synthesized sounds sound realistic and have clear main-order frequencies, making subsequent sound wave analysis more intuitive. The particle-synthesized sounds are coherent and free of stepped sounds under any vehicle conditions. The specific technical solution is as follows:

[0006] A method for synthesizing active car sounds, comprising the following specific steps:

[0007] Step 1: Extract sound particles and create corresponding sound particles according to the operating conditions of the car engine to form a sound particle library;

[0008] Step 2: Use periodic interpolation algorithm to synthesize the sound data to be played;

[0009] Step 3: Play the sound data without delay.

[0010] As a preferred embodiment: the step 1 is specifically as follows:

[0011] 1.1. Collect the original sound of the target vehicle;

[0012] 1.2. Obtain the main order frequency. Use a bandpass filter to extract the main order frequency of the original audio separately.

[0013] 1.3. Division of main order frequency zero crossing point;

[0014] 1.4. Particle segmentation: divide the original audio according to the zero-crossing position and time, and save each divided segment as an audio particle Pie i ;

[0015] 1.5. Determine the overlapping part and apply the Over-lapped principle to i+1 The first part of the data is spliced ​​in Pie i At the end of , ensure that the front and back sound particles can transition smoothly, multiply the spliced ​​sound particles with the window function and save them;

[0016] 1.6. Determine the rotation speed and index value corresponding to the sound particles.

[0017] As an example, the step 1.3, dividing the main order frequency by zero crossing point, is specifically as follows:

[0018] According to the main order frequency obtained, it can be seen as a sine curve y , as shown in formula (1), where A is the amplitude, ω is the frequency

[0019] (1)

[0020] Will y Cut the cycle into points and save the location of each zero crossing point. i , its purpose is to ensure that the engine sound is played in a full cycle.

[0021] As a preference: the step 1.6 specifically includes determining the speed value and index value index corresponding to the sound particle i , take the middle value of the frequency of each main order segment divided in step 1.4, and convert it into the rotation speed according to formula (2), where n is the number of engine cylinders, r is the engine speed per minute, fundamental frequency f e The corresponding harmonics are called the main orders of the engine sound

[0022] (2)

[0023] Then match the speed value with the sound particles in step 1.5 one by one and sort each particle by speed and mark the index value index i .

[0024] As a preference: the step 2 is specifically as follows:

[0025] 2.1. Obtaining first vehicle condition input information;

[0026] 2.2, according to the index value index i , get the corresponding sound particle Pie i , the audio data Apart_Pie to be played in the current time period delta_time i , the remaining audio data Rest_Pie i and the data to be superimposed Lap_Pie i ;

[0027] 2.3. Play separated audio particles Apart_Pie i , and store the speed value Rpm i , Rest_Pie i and Lap_Pie i ;

[0028] 2.4. Get the second vehicle condition input information Rpm i+1 , according to Rpm i Determine whether periodic interpolation is required. If not, repeat steps 2.1 to 2.3. If required, proceed to the next step.

[0029] 2.5. Perform sound period interpolation and extract Rpm i Corresponding Pie i , and Pie i+1 , since it is mentioned in the particle extraction that it is a quasi-periodic cut, so Pie i and Pie i+1 They are all complete periodic signals and are considered to have stable frequencies, so they are equivalent to:

[0030] (3)

[0031] (4)

[0032] Pie i and Pie i+1 Extract amplitude separately A , frequency ω, according to the extracted A and ω, and according to Rpm i and Rpm i+1 Calculate the scale factor α , and then useα Find the target amplitude A inter , frequency ω inter

[0033] (5)

[0034] (6)

[0035] Reuse A inter ,ω inter Constructing a periodic function of the target y inter , and capture the entire cycle and save it as Pie 00 ;

[0036] The Pie 00 Only the main order sound is included, and Pie i Perform band-stop filtering to obtain Pie 01 , the obtained Pie 01 And the newly obtained Pie 00 Synthesize a new Pie i+1 ;

[0037] 2.6、Calculate the current speed refresh time delta_time and the audio data to be played Apart_Pie i+1 , the remaining audio data Rest_Pie i+1 and the data to be superimposed Lap_Pie i+1 ;

[0038] 2.7. Repeat step 2.4. For the periodic interpolation algorithm involved in step 2.4, in order to ensure the continuity of its phase, it is necessary to use a zero-phase filter to obtain y inter Filtering is performed to ensure phase continuity.

[0039] As a preference: the step three is specifically as follows:

[0040] 3.1、Input the speed Rpmi and get the sound clip Apart_Pie to be played through the synthesis algorithm i ;

[0041] 3.2 Play Apart_Pie i ;

[0042] 3.3, Speed ​​refresh time delta_time;

[0043] 3.4 Input speed Rpm i+1 , get the sound clip to be played Apart_Pie through the synthesis algorithm i+1；

[0044] 3.5. Play Apart_Pie i+1 ;

[0045] 3.6. Repeat steps 3.4 to 3.5.

[0046] The beneficial effects of the present invention are as follows: the automobile sound synthesis method mentioned in the present invention adopts a periodic interpolation particle synthesis algorithm, which solves the problem of "step sound" in previous particle algorithms, increases the listening comfort, reduces the difficulty of obtaining the original sound of the target vehicle model in the particle algorithm, improves the accuracy of sound synthesis, ensures the sound continuity and smoothness of the automobile sound under any vehicle condition, and the continuity of each order line of the sound during software analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 This is a flow chart of the algorithm for sound particle synthesis and extraction in the present invention.

[0048] Figure 2 This is a flow chart of the sound synthesis algorithm in the present invention.

[0049] Figure 3 This is a flow chart of the sound playback process in the present invention. DETAILED DESCRIPTION

[0050] The preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more precise definition of the protection scope of the present invention.

[0051] like Figure 1 、 Figure 2 and Figure 3 As shown: A method for synthesizing active car sounds, the specific steps are:

[0052] First is the sound particle extraction method:

[0053] Step 1: Collect the original sound of the target vehicle. For the target vehicle model, use relevant recording equipment to obtain the car engine sound. The operating conditions of the car engine include but are not limited to "acceleration", "deceleration" and "constant speed". In order to increase the accuracy of the sound particles, the duration of the sound should be as long as possible. The acquisition of the sound here should ensure that the state of the car working conditions such as vehicle speed / rotation speed changes smoothly and monotonically, and the rate of change of the car working conditions should be coordinated with the total duration of the sound. There is no mandatory requirement for the collection position of the microphone or other recording equipment for sound acquisition, but it must meet the existence of a fixed transfer function at the location and the engine sound must be greater than other noises such as tire noise or wind noise;

[0054] Step 2: Obtain the main order frequency. Use a bandpass filter (bandpass filters include but are not limited to Butterworth bandpass filters, Gaussian bandpass filters, and elliptical bandpass filters. Here, the Butterworth bandpass filter is used as an example) to extract the main order frequency of the original audio. Here, a 6th-order Butterworth bandpass filter with a filter bandwidth of 4Hz is selected.

[0055] Step 3: The main order frequency is divided by zero crossing. The main order frequency obtained in step 2 is approximately regarded as a sine curve y, as shown in formula (1), where A is the amplitude and ω is the frequency

[0056] (1)

[0057] Cut y into cycles and save the position of each zero-crossing point i to ensure that the engine sound is played in a full cycle.

[0058] Step 4: Particle segmentation. Divide the original audio according to the zero-crossing position pointi-pointi+1, and save each segment as an audio particle Piei;

[0059] Step 5: Determine the overlapped part. Using the Overlapped principle, splice the front part of Piei+1 to the end of Piei to ensure a smooth transition between the front and back sound particles. Multiply the spliced ​​sound particles with the window function and save them (the window function here includes but is not limited to Hanning window, Hamming window, Blackman window, etc.);

[0060] Step 6: Determine the speed value and index value indexi corresponding to the sound particle. Take the middle value of the frequency of each main order segment divided in step 4 and convert it into speed according to formula (2), where n is the number of engine cylinders, r is the engine speed per minute, and the harmonic corresponding to the fundamental frequency fe is called the main order of the engine sound.

[0061] (2)

[0062] Then, the rotation speed values ​​are matched one by one with the sound particles in step 5, and each particle is sorted by the rotation speed and marked with the index value indexi.

[0063] The extracted sound particles contain their own sound information, and due to quasi-periodic cutting, the phases of any two audio particles are continuous. Based on this, the following sound synthesis method is proposed:

[0064] Step 1: Get vehicle condition input information 1, such as speed Rpmi (vehicle condition information includes engine speed, vehicle speed, etc., here we take speed as an example), and match the corresponding index value indexi in the audio particle library according to Rpmi;

[0065] Step 2: According to indexi, obtain the corresponding sound particle Piei, the audio data Apart_Piei to be played in the current time period delta_time, the remaining audio data Rest_Piei and the data to be superimposed Lap_Piei;

[0066] Step 3: Play Apart_Piei and store Rpmi, Rest_Piei and Lap_Piei;

[0067] Step 4: Obtain the second vehicle condition input information Rpmi+1, and determine whether periodic interpolation is required based on Rpmi. If not, repeat steps 1 to 3; if required, proceed to step 5.

[0068] Step 5: Perform sound period interpolation. Extract Piei and Piei+1 corresponding to Rpmi. Since quasi-periodic cutting is mentioned in particle extraction, Piei and Piei+1 are both complete periodic signals and are approximately regarded as stable in frequency, so they can be approximately equivalent to:

[0069] (3)

[0070] (4)

[0071] Piei and Piei+1 extract amplitude A and frequency ω respectively. According to the extracted A and ω, and according to Rpmi and Rpmi+1, the proportional factor α is calculated, and then the target amplitude Ainter and frequency ωinter are obtained by using α.

[0072] (5)

[0073] (6)

[0074] Then use Ainter and ωinter to construct the target periodic function yinter, and intercept the entire period and save it as Pie00;

[0075] The Pie00 only contains the main order sound, and Piei needs to be band-stop filtered to obtain Pie01, and the obtained Pie01 and the newly obtained Pie00 are synthesized into a new Piei+1;

[0076] Step 6: Calculate the audio data Apart_Piei+1 to be played at the current delta_time, the remaining audio data Rest_Piei+1, and the data to be superimposed Lap_Piei+1;

[0077] Step 7: Repeat step 4.

[0078] In particular, for the periodic interpolation algorithm involved in step 4, in order to ensure the continuity of its phase, it is necessary to filter the obtained yinter using a zero-phase filter to ensure the continuity of the phase.

[0079] For the synthesized sound data to be played, in order to ensure that the sound can be played without delay, a playback process is proposed.

[0080] When a speed signal is input, a file to be played with a delta_time duration is generated at the same time. Considering the calculation time of the algorithm and the duration of the sound particles at low speeds, the following playback process steps are required.

[0081] Step 1: Input the speed Rpmi and obtain the sound segment Apart_Pie to be played through the synthesis algorithm i ;

[0082] Step 2: Play Apart_Pie i ;

[0083] Step 3: Refresh the speed time delta_time;

[0084] Step 4: Input the speed Rpm i+1 , get the sound clip to be played Apart_Pie through the synthesis algorithm i+1；

[0085] Step 5: Play Apart_Pie i+1 ;

[0086] Step 6. Repeat steps 3.4 to 3.5.

[0087] Adding delta_time playback delay to the playback process can ensure that the synthesized sound can be played without delay.

Claims

1. A method for synthesizing active automobile sound, characterized in that: The specific steps used are: Step 1: Extract sound particles and create corresponding sound particles according to the operating conditions of the car engine to form a sound particle library; 1.

1. Collect the original sound of the target vehicle; 1.

2. Obtain the main order frequency. Use a bandpass filter to extract the main order frequency of the original audio separately. 1.

3. Division of main order frequency zero crossing point; The main order frequency is divided into zero-crossing points, specifically, According to the main order frequency obtained, it can be seen as a sine curve y , as shown in formula (1), where A is the amplitude, ω is the frequency （1） Will y Cut the cycle into points and save the location of each zero crossing point. i , its purpose is to ensure that the engine sound is played in a full cycle; 1.

4. Particle segmentation: divide the original audio according to the zero-crossing position and time, and save each divided segment as an audio particle Pie i ; 1.

5. Determine the overlapping part and apply the Over-lapped principle to i+1 The first part of the data is spliced ​​in Pie i At the end of , ensure that the front and back sound particles can transition smoothly, multiply the spliced ​​sound particles with the window function and save them; 1.

6. Determine the speed value and index value corresponding to the sound particle; Step 2: Use periodic interpolation algorithm to synthesize the sound data to be played; 2.

1. Obtaining first vehicle condition input information; 2.2, according to the index value index i , get the corresponding sound particle Pie i , the audio data Apart_Pie to be played in the current time period delta_time i , the remaining audio data Rest_Pie i and the data to be superimposed Lap_Pie i ; 2.

3. Play separated audio particles Apart_Pie i , and store Rpm i , Rest_Pie i and Lap_Pie i ; 2.

4. Get the second vehicle condition input information Rpm i+1 , according to Rpm i Determine whether periodic interpolation is required. If not, repeat steps 2.1 to 2.

3. If required, proceed to the next step. 2.

5. Perform sound cycle interpolation and extract Rpm i Corresponding Pie i , and Pie i+1 , since it is mentioned in the particle extraction that it is a quasi-periodic cut, so Pie i and Pie i+1 They are all complete periodic signals and are considered to have stable frequencies, so they are equivalent to: （3） （4） Pie i and Pie i+1 Extract amplitude separately A , frequency ω, according to the extracted A and ω, and according to Rpm i and Rpm i+1 Calculate the scale factor α , and then use α Find the target amplitude A inter , frequency ω inter （5） （6） Reuse A inter ,ω inter Constructing a periodic function of the target y inter , and capture the entire cycle and save it as Pie 00 ; The Pie 00 Only the main order sound is included, and Pie i Perform band-stop filtering to obtain Pie 01 , the obtained Pie 01 And the newly obtained Pie 00 Synthesize a new Pie i+1 ; 2.6、Calculate the current speed refresh time delta_time and the audio data to be played Apart_Pie i+1 , the remaining audio data Rest_Pie i+1 and the data to be superimposed Lap_Pie i+1 ; 2.

7. Repeat step 2.

4. For the periodic interpolation algorithm involved in step 2.4, in order to ensure the continuity of its phase, it is necessary to use a zero-phase filter to obtain y inter Filtering is performed to ensure phase continuity; Step 3: Play the sound data without delay; 3.1、Input the speed Rpmi and get the sound clip Apart_Pie to be played through the synthesis algorithm i ; 3.2 Play Apart_Pie i ; 3.3, Speed ​​refresh time delta_time; 3.4 Input speed Rpm i+1 , get the sound clip to be played Apart_Pie through the synthesis algorithm i+1； 3.

5. Play Apart_Pie i+1 ; 3.

6. Repeat steps 3.4 to 3.

5.

2. The method for synthesizing active automobile sound according to claim 1, wherein: The step 1.6 specifically includes determining the speed value Rpm corresponding to the sound particle. i and index value index i , take the middle value of the frequency of each main order segment divided in step 1.4, and convert it into the rotation speed according to formula (2), where n is the number of engine cylinders, r is the engine speed per minute, fundamental frequency f e The corresponding harmonics are called the main orders of the engine sound （2） Then match the speed value with the sound particles in step 1.5 one by one and sort each particle by speed and mark the index value index i .

Images