TW201816774A - Audio processing method and non-transitory computer readable medium - Google Patents

Abstract

An audio processing method is disclosed. The audio processing method includes: splitting an audio file into a plurality of audio sections; analyzing a first lowest energy value within a frequency spectrum waveform of a first audio section of the audio sections; comparing the first lowest energy value with a predetermined energy value and assigning the higher one as a first noise floor; generating a first processed audio section according to the first noise floor and the first audio section; compressing the first processed audio section to generate a compressed audio section; and sending the compressed audio section to an audio playback apparatus.

Description

 Audio processing method and non-transitory computer readable medium  

The present disclosure relates to an audio processing method, and more particularly to an audio processing method and a non-transitory computer readable medium for compressing an audio file.

Traditionally, if an audio file is to be transmitted to an audio playback device via a wireless transmission protocol such as Bluetooth that supports only a low-bandwidth, it is necessary to use a distortion/lossy compression method such as an MP3 format to greatly reduce the amount of data, but the distortion is compressed. The method may seriously cause the loss of low frequency and high frequency sound in the audio file, or reduce the original rich frequency or volume change, which greatly reduces the audio quality.

In addition, general compression techniques usually involve a large number of operations such as converting audio files between the time domain and the frequency domain. However, small-sized playback devices such as Bluetooth headsets, Bluetooth speakers, etc. usually only have low-processing microprocessors, so the solution is executed. When compressing audio files, these small broadcasters will take a long time to process and cannot be played instantly.

An audio processing method is proposed in one of the technical aspects of the present disclosure. The audio processing method first divides the audio file into a plurality of audio segments. The first lowest energy value in the spectral waveform of the first audio segment of the audio segments is then analyzed. The first lowest energy value is compared with a predetermined energy value, and the higher one is used as the first noise floor. Generating a first processed audio segment based on the first noise floor and the first audio segment. The first processed audio segment is compressed to produce a compressed audio segment. And transmitting the compressed audio segment to the audio playback device.

A non-transitory computer readable medium is presented in another aspect of the disclosure. The non-transitory computer readable medium stores a plurality of instructions. When the plurality of instructions are executed by the processing unit, the method performs: dividing the audio file into a plurality of audio segments, and processing one of the audio segments according to the following steps: analyzing the The lowest energy value of the spectral waveform of one of the audio segments; comparing the lowest energy value with a predetermined energy value, and using the higher one as the noise floor; generating the classic according to the noise floor and the one audio segment Processing the audio segment; compressing the processed audio segment to generate a compressed audio segment; and transmitting the compressed audio segment to the audio playback device.

A non-transitory computer readable medium is presented in yet another aspect of the disclosure. The non-transitory computer readable medium stores a plurality of instructions for restoring the compressed audio section in the compressed audio file. When the plurality of instructions are executed by the processing unit, performing: decompressing the compressed audio section to obtain a decompressed audio zone Segment; and multiplying each sample value in the decompressed audio segment by a discard value. The discard value is related to the original noise floor of the original audio segment corresponding to the compressed audio segment.

Through the teachings of this disclosure, audio files can be transmitted via a low frequency wide transmission protocol. Since the audio file is processed in a distortion-free compression format, which does not involve, for example, conversion between the time domain and the frequency domain, even if the audio playback device has only a low computing power processor, the audio file can be quickly decompressed. For instant playback.

100, 400, 500‧‧‧ audio processing methods

S102, S104, S106, S108, S109‧‧‧ steps

S110, S111, S112, S114, S115‧‧‧ steps

S116, S118, S119, S120‧‧ steps

L11, L12, L13, L14‧‧‧ energy values

FIG. 1 is a flow chart of an audio processing method according to an embodiment of the disclosure.

2A-2C are spectrum waveform diagrams of an embodiment of the present disclosure.

3A-3C are time-domain waveform diagrams of an embodiment of the present disclosure.

FIG. 4 is a flow chart of an audio processing method according to an embodiment of the disclosure.

FIG. 5 is a flow chart of an audio processing method according to an embodiment of the disclosure.

Figure 6 is a graph showing the function of one embodiment of the present disclosure.

The following detailed description of the embodiments of the present invention is intended to be illustrative of the invention, and is not intended to limit the invention, and the description of structural operation is not intended to limit the order of execution, any The means for re-combining the components, resulting in equal functionality, are within the scope of the present disclosure. Moreover, the drawings are only for illustrative purposes and are not drawn in their true dimensions.

The terms used in the entire specification and the scope of the patent application, unless otherwise specified, generally have the ordinary meaning of each term used in the field, the content disclosed herein, and the particular content. Certain terms used to describe the disclosure are discussed below or elsewhere in this specification to provide additional guidance to those skilled in the art in the description of the disclosure.

FIG. 1 is a flow chart showing an audio processing method 2 according to an embodiment of the disclosure. The audio processing method 100 is configured to compress the audio file and send the compressed audio file to the playback device for playback. Preferably, when the audio file is large, the audio processing method 100 can divide the audio file into multiple audio segments. Individual processing is performed for each audio segment. Audio files can be segmented according to any rules, such as length of time, number of samples, and/or file size. The audio processing method 100 processes each audio segment according to the chronological order of the audio content, and the content of each audio segment has the same or different length of time, the number of sampling points, and/or the file size. The file is not restricted.

The audio processing method 100 includes steps S102 to S120. Here, steps S102 to S114 are executed by, for example, a device having a higher arithmetic processing capability such as a computer, and steps S116 to S120 are executed by, for example, a device having a lower arithmetic processing capability such as a Bluetooth device. For example, the above-mentioned arithmetic processing capability refers to operational parameters such as the clock rate of the processor, the performance of the processor, the floating point computing capability, the bit width, and the capacity of the memory. For example, a device with higher arithmetic processing capability may include Audio systems, smart phones, tablets, portable music players, etc., lower computing power can include Bluetooth headsets, Bluetooth speakers and so on.

The first audio segment of the plurality of audio segments in the audio file will first be processed through steps S102-S120. After the first audio segment is processed by the audio processing method 100, the second audio segment is processed through steps S102 to S120, and after the second audio segment is processed, the next audio segment is continuously executed. In other words, each audio segment is processed through steps S102-S120 in sequence until the entire audio file is processed. Steps S102~S110 are all pre-processing steps before compressing the audio segment. In the following, only the first audio segment and the second audio segment are taken as an example to simplify the description.

In step S102, the first audio segment is converted from the time domain data to the data (spectrum) expressed in the frequency domain, and the conversion may be performed by, for example, Fast Fourier Transform (FFT) or other similar algorithms. . The data is a sampling point in the time domain or the frequency domain and corresponding sampling value data. For the result of the conversion, the spectrum waveform of the first audio segment of one embodiment of the disclosure may be referred to in FIG. 2A. In Fig. 2A, the horizontal axis coordinate unit is frequency (Hz), and the vertical axis is volume/energy (dB).

Next, in step S104, the lowest energy value in the spectral waveform of the first audio segment is analyzed. The purpose of this step is to calculate the amount of data occupied by unnecessary system noise. For example, the audio output usually contains system-specific noise at each point in time. This system noise is generally called a noise reference or a noise floor. The noise floor is undesired noise, affecting the signal-to-noise ratio (SNR), and the noise ratio is related to the quality of the audio. The noise floor is particularly noticeable in the silent phase of the audio, which also limits the dynamic range of the audio (the ratio of the strongest volume to the weakest volume). Therefore, the removal of the amount of data occupied by the system noise, in addition to reducing the file size, improving the compression capacity of subsequent compression processing, also improves the quality of the audio (increased SNR).

In step S104, the analyzed lowest energy value is taken as the first lowest energy value. In the spectrum of the embodiment of Fig. 2A, the first lowest energy value is at the energy value L11, which is about -130 dB. In general, high-frequency data usually has a lower energy in a piece of audio content. It should be noted that the maximum frequency range of the sound that can be perceived by the human ear is about 20 Hz to 20 kHz, but the perception of sound above 15 kHz is already weak. Therefore, in popular music records or certain audio files, for example, the record production company will first remove the higher frequency (for example, 15KHz or higher) audio content in the audio file to reduce the file size, as shown in FIG. 2B. FIG. 2B is a diagram showing a spectrum waveform in which audio content of a frequency higher than 15 kHz is removed in an embodiment of the disclosed document. That is to say, there is no useful information in the audio part above the high frequency of 15KHz, and only the useless information (noise) remains. In Fig. 2B, the horizontal axis coordinate unit is frequency (Hz), and the vertical axis is volume/energy (dB).

In the embodiment of FIG. 2B, the first lowest energy value analyzed through step S104 is approximately at 45 KHz, which corresponds to the energy value L12 (-120 dB) indicated in the figure. However, in fact, in the embodiment of FIG. 2B, there is no valid audio file content in the audio segment above 15 kHz (which has been removed by the record production company at the factory), that is, the portion ranging from 15 kHz to 45 kHz is also The amount of data possessed by unnecessary system noise. Therefore, in step S106 of the audio processing method 100, the first lowest energy value analyzed in step S104 is compared with a preset energy value, and the higher one is used as the first noise floor. Among them, in the disclosure document, the data below the energy value corresponding to the first noise floor is regarded as so-called noise. For example, if the lowest energy value analyzed in step S104 is lower than the preset energy value, the preset energy value is used as the noise floor, and when the lowest energy value analyzed is higher than the preset energy value, the lowest energy is used. The value is used as the noise floor.

In the embodiment of FIG. 2B, the preset energy value corresponds to the energy value L13 (eg, -85 dB). The preset energy value can also be set by the user, and the disclosure document is not limited. In this example, the preset energy value (-85dB) is higher than the lowest energy value (-120dB), so the preset energy value is -85dB as the first noise floor, and below the -85dB first noise floor energy value. The information is considered as noise.

The preset energy value -85dB corresponds to the 15KHz frequency in Figure 2B. Therefore, the portion of the range of 15KHz to 45KHz (the lowest energy value corresponding to the frequency) can also be classified as noise data by setting the preset energy value. It will not be misplaced and limit the ability of subsequent file compression. Briefly, by step S106, the noise/necessary data closer to the actual audio can be calculated.

In another case, if the measured minimum energy value is higher than the preset energy value, the measured lowest energy value is used as the first noise floor. Please refer to FIG. 2C. FIG. 2C is a diagram showing a waveform waveform of an embodiment of the disclosed document. In the spectrum of the audio block of FIG. 2C, the lowest energy value L14 is about -78 dB, which is higher than the preset energy value (-85 dB), so the lowest energy value L14 is used as the first noise floor. The measured minimum energy value is used as the noise floor, and the portion below the noise floor is classified as noise data, so that the noise floor will be set with the lowest energy value of the audio content. Does not fix the noise floor to the preset energy value.

Next, in step S108, the first discarded value is generated according to the data in the time domain waveform of the first audio segment that is lower than the first noise bottom energy value. The first discard value is used for further processing with the first audio segment to generate a first processed audio segment. In detail, step S108 calculates a time domain amplitude by performing a Root Mean Square (RMS) operation on a sample value of a time domain waveform of the first audio segment that is lower than a sampling point of the first noise floor. Amplitude), and use this amplitude as the first discard value. Then, in step S110, each initial sample value in the first audio segment is divided by the first discarded value, and after the decimal point is unconditionally rounded to the integer bit, the first processed audio segment is generated. For example, the unconditional rounding off of the decimal point mentioned above can be done by the floor function.

Assume that the first audio segment is in 24-bit/96KHz format, and the data range that can be presented by 24bit has 8388608 different intensity levels, for example, it can be used to represent the value interval -3888818~-1, or can be used to represent the value. Interval 0~8388607, or other set value range. The following example is described using the numerical interval 0~8388607.

When the first audio segment is in the time domain, the initial sampling value of one of the sampling points is the maximum value of 8388607 which can be presented in the 24-bit format, and the first rejection value is assumed to be 1000. In step S110, the value of the sample point 8388607 is divided by 1000 to obtain 8386.607, and is taken to the whole number by the rounding function to obtain a new sample value of 8388. That is, after the sampling point whose initial sampling value is 8388607 in the original first audio section is processed in step S110, the sampling value of the same sampling point in the corresponding first processed audio section is 8388.

Therefore, the audio of the 24-bit/96KHz format originally uses the data amount of 24 bits to store the data of each sampling point, and after the pre-compression pre-processing step through steps S102-S110, the maximum initial sampling value corresponds to the new maximum sampling value. For 8388 (between 2 ¹³ and 2 ¹⁴ ), each sample point can be stored using only 15 bits of data. Thereby, the ability to subsequently compress audio can be greatly improved. It should be noted that the traditional method for the noise floor is processed according to the number of bits. For example, when the first discard value is 1000, since 1000 is between 2 ⁹ and 2 ¹⁰ , only 2 can be discarded. ⁹ (= 512) of the amount of data, wasted 1000-512 = 488 discarded value data. That is to say, the traditional practice may still retain unnecessary part of the noise, which leads to the decline of subsequent compression capabilities.

In the embodiment, when the sampled value of a sampling point is lower than the first discarded value, the new sampled value will be zero. For example, assume that the sampling value of one of the sampling points of the first audio segment in the time domain is 900 (below the assumed first discarding value of 1000). After the processing in step S110, the value 900 of the sampling point is divided by 1000 to obtain 0.9, and the integer value is obtained by taking the rounding function to obtain an integer value of 0. That is, when the initial sampling value in the original first audio segment is lower than the first discarded value, the new sampling value in the corresponding first processed audio segment is 0 after the processing in step S110.

Next, step S112 compresses the first processed audio segment to generate a compressed audio segment. In detail, since the file size of the first audio segment has been greatly reduced by the pre-processing steps of steps S102 to S110, step S112 can compress the first processed audio segment using the distortion-free compression format. There is no need to pass the distortion compression format to improve compression. In this embodiment, the distortion-free compression format is, for example, Free Lossless Audio Codec (FLAC). With FLAC compression technology, the sampling point of the lowest sampled value (for example, 0) in the first processed audio segment is discarded first to improve the compression capability, and the sampling point of the lowest sampling value is restored to restore the original sampling after being decompressed. rate. Wherein, if the first audio segment is directly compressed without being preprocessed in steps S102 to S110, the compression ratio (the ratio of the compressed size to the size before compression) that FLAC compression can provide is about 70%-80. %, and after the pre-processing of steps S102-S110, compression is performed, and the compression ratio can reach 20% to 15%.

After the first processed audio segment is compressed to generate a compressed audio segment, step S114 sends the compressed audio segment to an audio playback device, such as a Bluetooth headset or a Bluetooth speaker, for example, through a Bluetooth transmission, such as a Bluetooth headset or a Bluetooth speaker. The device of ability. In step S116, the audio playback device can decompress and restore the received compressed audio segment. Since the compressed audio segment is generated by processing without distortion compression (for example, FLAC), in the decompression process, it is only necessary to restore the sampling point of the lowest sample value that is removed during compression (ie, returning to the first processed audio) Section), without the need for additional complex and large numbers of operations such as inverse fast Fourier transforms.

After the decompression is restored, step S118 multiplies the sampled value of each sample point of the restored first processed audio segment by the first discarded value to restore the original audio format (for example, 24 bits). Then, in step S120, the restored audio is played in real time. Therefore, the audio processed by the audio processing method 100 can be quickly decompressed and restored by the audio playback device for instant playback.

In the embodiment, after the first audio segment is processed by the audio processing method 100, the second audio segment is then processed by the audio processing method 100. Step S102 first converts the time domain data of the second audio segment into a frequency spectrum. Step S104 analyzes a second lowest energy value in the spectral waveform of the second audio segment. Step S106 compares the second lowest energy value with the preset energy value, and uses the higher one as the second noise floor. In step S108, the amplitude value (Amplitude) of the time domain is calculated by performing a root mean square (RMS) operation on the sampling value of the sampling point of the second audio segment in which the energy value is lower than the sampling point of the second noise floor. The amplitude magnitude is used as the second discard value and is processed with the second audio segment in step S110 to generate a second processed audio segment.

Next, step S112 is performed to compress the second processed audio segment and step S114 to send the compressed audio to the playback device, and the decompression and reduction process of steps S116 and S118 is performed, and finally the audio is played in step S120.

In one embodiment, the time domain waveform diagram of the audio segment processed by the audio processing method 100 is shown, for example, in FIGS. 3A-3C. In the 3A~3C diagram, the unit of the abscissa axis is time (t), and the unit of the ordinate axis is the intensity level, that is, the sample value. FIG. 3A is a diagram showing an original time domain waveform of an audio segment of an embodiment of the disclosed document. FIG. 3B is a time-domain waveform diagram of the processed audio segment generated by the pre-processing of steps S102-S110 in the audio segment of the embodiment of FIG. 3A. In this example, it is assumed that the discard value of 448 is calculated in step S108 to process the audio segment. The 3C is a time domain waveform diagram of the processed audio segment of FIG. 3B after the step S112 compression, the S114 transmission, and the S115~118 decompression and reduction process. As can be seen from Figures 3A and 3C, the audio segments processed by the audio processing method 100 are not significantly distorted.

In an embodiment of the disclosure, the audio processing method may further include step S109 and step S115, as shown in FIG. FIG. 4 is a flow chart of an audio processing method 400 according to an embodiment of the disclosure. The audio processing method 400 includes steps S102, S104, S106, S108, S109, S110, S112, S114, S115, S116, S118, S120, wherein steps S102~S108, S110~S114, S116~S120 are the same as the audio processing method 100, Please refer to the description of the relevant paragraphs above, and the details are not repeated here. After the first discard value is generated in step S108, the first discard value is multiplied by an adjustment coefficient in step S109. The adjustment coefficient can be customized by the user to control and adjust the quality of the audio file generated by the subsequent processing steps.

In detail, if the user can judge that the audio file does not need too high quality, the user can choose to increase the first discard value, thereby increasing the amount of data to be discarded, thereby reducing the size of the audio file, and the subsequent compression capability can be further improved. For example, assuming that the first discard value is 1000 and the adjustment coefficient is 16, in step S109, the first discard value 1000 is multiplied by an adjustment factor of 16, and the product is a new discard value of 16000, that is, the discard value is increased. Then, in step S110, each initial sample value in the first audio segment is divided by the new discard value, and processed by the lower rounding function to generate a first processed audio segment. Then, the first processed audio segment is compressed to generate a compressed audio segment via step S112, and then sent to the audio playback device in step S114.

In step S115, the transmission bandwidth of the transmitted compressed audio segment is calculated, and if the transmission bandwidth is greater than a preset value, the adjustment coefficient of the next audio segment (second audio segment) is increased. In general, in order for Bluetooth to transmit data stably, the bandwidth usually needs to be between 1 and 1.5 Mbps or less. In this embodiment, the preset value is set to 660 Kbps. When the bandwidth of the transmitted compressed audio segment is greater than a preset value, the adjustment coefficient of the second audio segment is automatically increased, thereby increasing the discarding value to improve the compression capability. Due to the increase of the adjustment factor, the transmission bandwidth of the subsequently compressed audio segment will conform to the condition of stable transmission (less than 660 Kbps).

It should be understood that when the transmission bandwidth is much smaller than the preset value, the adjustment coefficient of the second audio segment may also be lowered to increase the bandwidth. The value of the adjustment coefficient may be an integer/non-integer or even a functional formula, and the disclosure is not limited. In an embodiment, the system or the user may also establish an adjustment coefficient table in advance. The adjustment coefficient table includes a plurality of different adjustment coefficients, so in step S115, when the transmission bandwidth is greater than or far less than the preset value, the audio processing method 400 can automatically select a larger or smaller adjustment coefficient in the adjustment coefficient table. To process the next audio segment.

In another embodiment of the disclosure, the audio processing method may further include steps S111 and S119. The flowchart of the audio processing method 500 according to an embodiment of the disclosure is shown in FIG. The audio processing method 500 includes steps S102, S104, S106, S108, S111, S112, S114, S116, S119, S120, wherein steps S102~S108, S112~S116, S120 are the same as the audio processing method 100, please refer to the foregoing related paragraphs. This will not be repeated here. In step S111, the first discard value generated in step S108 is dynamically adjusted according to the size of each initial sample value in the first audio segment to further generate a processed audio segment. That is, the sample values of each sampling point are adjusted according to the respective first discarded values. The first discard value and each initial sample value of the first audio segment are converted by a nonlinear companding method to correspondingly adjust each initial sample value and generate a new sample value.

In an embodiment, the nonlinear companding method may be, for example, a mu-law encoding. In Mu-law coding, the initial sample value interval is corresponding to the interval where the maximum value is 1 and the minimum value is -1, that is, the sampled value is divided by the maximum value. The Mu-law function is as follows:

Where x is the sampled value, μ is the discarded value, and sign(x) is the sign function. When x is greater than 0, then sign(x)=1; when x is 0, then sign(x)=0; When x is less than 0, sign(x) = -1. The value of mu(x) is set between 1 and -1. Therefore, the calculated value of mu(x) must be multiplied by the number of bits of the converted audio format to obtain the actual corresponding sample value. For the relationship between the Mu-law encoding function mu(x) and the sampled value x, see the Mu-law encoding function graph of one embodiment of the present disclosure shown in FIG. In Fig. 5, the horizontal axis represents the value of x and the vertical axis represents mu (x).

For example, if the first audio segment is in the 16bit/44.1KHz format, when the discard value μ is 255, the data amount of the first audio segment will be converted to 8 bits after the processing in step S111. If there is a sampling point with a sampling value of 33, after Mu-law encoding conversion, mu (33/32768)=0.0412 can be obtained, and the data amount of the first audio segment is converted into 8 bits after processing, so 0.0412 times is multiplied. On top of 2 ⁷ (=128), and after the rounding of the whole function, get 5. That is to say, the sampling point with a sampling value of 33 corresponds to the sampling value 5 in the 8-bit format after being encoded by Mu-law. Or, there is another sampling point with a sampling value of 32178. After Mu-law encoding conversion, mu (32178/32768)=0.9967, then 0.9967 multiplied by 128, and the following rounding function is obtained to obtain 127. . That is, the sampling point with the sampling value of 32178 is encoded by Mu-law, and corresponds to the sampling value 127 in the 8-bit format.

By discarding the value by Mu-law encoding, even the sample points with smaller amplitudes can be retained, so that the dynamic range of the audio segment can be saved, so the audio quality is not lost too much due to the processing of the noise. It should be understood that the audio processing method 500 may use different nonlinear companding techniques according to actual applications. The Mu-law encoding is only described as a preferred embodiment, but is not intended to limit the disclosure.

After the step S111 is completed, the step S112 is performed to compress the file and the step S114 is to send the compressed audio segment to the audio playback device. The audio playback device decompresses the compressed audio segment to restore the original processed audio segment through step S116. Next, in step S119, reverse Mu-law processing is performed to restore to the audio section of the original audio format. Among them, the inverse μ-law function is as follows:

Taking the sampling point with the above sample value of 33 as the example, the sample value 5 corresponding to the 8-bit format is obtained by Mu-law coding, and the sample value 5 is placed in the inverse Mu-law function to obtain mu_inverse(5/128)= 0.00094846, because the data volume of the original first audio segment is 16 bits, multiply 0.00094486 by 2 ¹⁵ (=32768), and the decimal point is unconditionally carried to obtain 32, which is only about 3% error with the original sample value 33. For example, the unconditional carry of the decimal point mentioned above can be done by a ceiling function. Taking the sample point with the sample value of 32178 as described above and the sample value 127 corresponding to the 8-bit format after the Mu-law coding, the sample value 127 is added to the inverse Mu-law function, and mu_inverse (127/128) is obtained. = 0.9574, multiplying 0.9574 by 2 ¹⁵ and passing the upper rounding function to obtain 31373, which is only about 2.5% error from the original sample value 32178.

In an embodiment of the present disclosure, the steps in the audio processing methods 100, 400, and 500 may be integrated or changed. For example, an audio processing method may also integrate step S109 including the audio processing method 400. S115, and steps S111 and S119 including the audio processing method 500. Specifically, the first discard value may be multiplied by the adjustment coefficient by step S109 of the audio processing method 400 to generate a new discard value, and then the first audio segment and the new discard value are substituted into step S111 of the audio processing method 500 to transmit the non-transparent value. The linear companding technique produces a first processed audio segment. Then, after compression and transmission, the transmission bandwidth of the transmitted compressed audio segment is calculated in step S115 to determine whether there is a need to increase the adjustment coefficient of the next audio segment.

In an embodiment of the disclosure, the audio processing method can be implemented by a non-transitory computer readable medium. The non-transitory computer readable medium stores a plurality of code instructions. When the plurality of code instructions are executed by the processing unit, the steps S102, S104, S106, S108, S109, and S110 of the audio processing methods 100, 400, and 500 can be performed. , S111, S112, S114, S115 or an integrated method of these steps. The non-transitory computer readable medium can be a computer, a mobile phone, or a standalone audio encoder, and the processing unit can be a processor or a system chip.

In another embodiment of the disclosure, another non-transitory computer readable medium stores a plurality of code instructions. When the plurality of code instructions are executed by the processing unit, the audio processing methods 100, 400, Steps S116, S118, S119, and S120 in 500. The other non-transitory computer readable medium can be an audio/video device such as a Bluetooth/wireless earphone, a speaker, a stereo, or a separate audio decoder, and the processing unit can be a microprocessor or a system chip.

Through the teachings of the disclosed documents, even if the audio file uses a high-resolution format of 24bit/96KHz, it can be transmitted after being compressed, and transmitted through a transmission standard of low transmission bandwidth such as Bluetooth, and can be quickly and instantaneously played on the audio playback device. Play.

Although the embodiments of the present invention have been disclosed as above, it is not intended to limit the present invention, and any person skilled in the art can make some modifications and retouchings without departing from the spirit and scope of the present invention. The scope is defined as defined in the scope of the patent application.

Claims (10)

 An audio processing method includes: dividing an audio file into a plurality of audio segments, and processing the first audio segment of the audio segments includes the following steps: analyzing one of the spectral waveforms of the first audio segment a first lowest energy value; comparing the first lowest energy value with a predetermined energy value, and using the higher one as a first noise floor; generating a first according to the first noise floor and the first audio segment a first processed audio segment; compressing the first processed audio segment to generate a compressed audio segment; and transmitting the compressed audio segment to an audio playback device.    The audio processing method of claim 1, wherein the step of generating the first processed audio segment further comprises: at least one of a time domain waveform of the first audio segment having an energy value lower than a first noise floor The sampled values of the sampling points are subjected to a root mean square operation to generate a first discarded value; and the initial sampled values in the first audio segment are divided by the first discarded value to generate the first processed audio segment.    The audio processing method of claim 1, wherein the step of generating the first processed audio segment further comprises: at least one of a time domain waveform of the first audio segment having an energy value lower than a first noise floor The sampled value of the sampling point is subjected to a root mean square operation to generate a first discarded value; and each initial sampled value is correspondingly adjusted according to the first discarded value and each initial sampled value in the first audio segment.    The audio processing method of claim 1, further comprising: analyzing a second lowest energy value of the spectrum waveform of the second audio segment, wherein the second audio segment is sent after the first audio segment; Comparing the second lowest energy value with the preset energy value, and using the higher one as a second noise floor; the energy value in the time domain waveform of the second audio segment is lower than the second noise floor The sampled value of the at least one sampling point is subjected to a root mean square operation to generate a second discarded value; and the second audio zone is adjusted when the bit rate of the compressed audio segment sent to the audio playback device is greater than a preset value The second discard value of the segment.    The audio processing method of claim 4, further comprising: when the bit rate of the compressed audio segment sent to the audio playback device is greater than the preset value, multiplying the second discarded value by an adjustment coefficient; The product of the second discard value and the adjustment factor adjusts a plurality of initial sample values of the second audio segment to generate a second processed audio segment.    The audio processing method of claim 1, wherein the audio playback device is a Bluetooth device, and transmitting the compressed audio segment to the audio playback device is transmitted via Bluetooth.    The audio processing method of claim 1, wherein the step of compressing the processed audio segment is distortionless compression.    A non-transitory computer readable medium storing a plurality of instructions, when the plurality of instructions are executed by a processing unit, performing: dividing an audio file into a plurality of audio segments, and processing the audio segments according to the following steps: An audio segment: analyzing a lowest energy value of the spectral waveform of the one audio segment; comparing the lowest energy value with a predetermined energy value, and using the higher one as a noise floor; according to the noise And the one of the audio segments to generate a processed audio segment; compressing the processed audio segment to generate a compressed audio segment; and transmitting the compressed audio segment to an audio playback device.    The non-transitory computer readable medium of claim 8, wherein the step of generating the processed audio segment further comprises: performing an energy value in a time domain waveform of one of the audio segments below the noise floor A sample value of at least one sample point is subjected to a root mean square operation to generate a discard value; and each initial sample value in the audio segment is divided by the discard value to generate a processed audio segment.    A non-transitory computer readable medium storing a plurality of instructions for restoring a compressed audio segment in a compressed audio file, and when the plurality of instructions are executed by a processing unit, performing: decoding the compressed audio segment Compressing to obtain a decompressed audio segment; and multiplying each sample value in the decompressed audio segment by a discard value; wherein the discarding value is an original noise floor of an original audio segment corresponding to the compressed audio segment Related.