JPH0736482A - Playback speed conversion method - Google Patents

Abstract

(57) [Abstract] [Object] The present invention relates to a reproduction speed conversion method of dividing audio data into a certain time period, and combining the data to convert a reproduction speed. The data with the smaller level difference between the data and the inverted data is joined to reduce the discontinuity at the joint,
The purpose is to reduce noise. [Constitution] To calculate the level difference between the data obtained by dividing the audio data into fixed time periods and the inverted data obtained by inverting this data, and joining the data with the smaller level difference, respectively. Repeat
It is configured to convert the reproduction speed of the audio data.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a reproduction speed conversion method in which audio data is divided into fixed time periods and the data is combined to convert the reproduction speed.

[0002]

2. Description of the Related Art Conventionally, in the case of changing the reproduction speed of digital audio data without changing the pitch, the digital audio data is divided at a fixed time period, and the divided data is connected or inserted. For example, in FIG.
The original digital audio data shown in (4) is divided at a fixed time period and the data is joined one by one to obtain the digital audio data of FIG. 7B, and the reproduction speed is doubled. On the contrary, when the reproduction speed is slowed down, for example, the original digital audio data of (a) of FIG. 8 is divided into a certain period of time, and the divided data is repeated to be the digital audio data of (b) of FIG. I was slowing it in half.

FIG. 7 shows a conventional reproduction speed conversion example (in the case of increasing the speed). FIG. 7A shows a state in which the original digital audio data is divided into fixed time periods.

FIG. 7B shows a state in which the divided data of FIG. 7A is joined one by one to double the reproduction speed. FIG. 8 shows a conventional reproduction speed conversion example (when the reproduction speed is slowed).

FIG. 8A shows a state in which the original digital audio data is divided into fixed time periods. FIG. 8B
8 shows a state in which the segmented data of FIG. 8A is repeated to reduce the reproduction speed by half.

[0006]

Although the conventional reproduction speed conversion method shown in FIGS. 7 and 8 has a simple processing content, a discontinuous portion is generated at the joined portion of the separated data. When listening to the digital audio data after changing the playback speed, noise was noticeable.

In order to suppress this noise, a low-pass filter or the like may be used, but there is a problem that the discontinuous portion is not effective because it contains all the frequencies, like the impulse.

The present invention solves these problems by
When joining delimited audio data, the data with the smaller level difference between the delimited data and the inverted data is joined to reduce the discontinuity at the joint and reduce noise. There is.

[0009]

[Means for Solving the Problems] Means for solving the problems will be described with reference to FIG. In FIG. 1, the data inversion processing unit 3 inverts data divided into fixed time periods.

The level difference calculation unit 4 stores data divided into fixed time periods and inverted data obtained by inverting this data,
The level difference from the data immediately before joining is calculated.

The level difference judging section 5 judges the data having the smaller level difference calculated by the level difference calculating section 4.

[0012]

According to the present invention, as shown in FIG. 1, the data inversion processing unit 3 inverts data divided into fixed time periods, and the level difference calculation unit 4 divides data into fixed intervals and inverted data. And a level difference from the data to be joined is calculated, and the level difference determination unit 5 determines the data having the smaller level difference calculated by the level difference calculation unit 4,
The determined data and the immediately preceding data are joined and output repeatedly to convert the reproduction speed of the audio data.

At this time, the data obtained by dividing the audio data into fixed time periods are joined one by one or a plurality of times to increase the reproduction speed. Further, regarding the data obtained by dividing the audio data into a certain time period, the same data is joined once or a plurality of times to reduce the reproduction speed.

Therefore, when the audio data divided into a fixed time period is combined, the data having the smaller level difference between the separated data and the inverted data is combined, so that the discontinuity of the connection portion becomes large. And noise can be reduced.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the construction and operation of an embodiment of the present invention will be described in detail with reference to FIGS.

FIG. 1 shows a block diagram of an embodiment of the present invention.
FIG. 1A shows a configuration diagram. In FIG. 1A, audio data 1 is digital audio data before conversion of reproduction speed.

The data reading unit 2 reads the data divided from the audio data 1 in a fixed time period.
The data inversion processing unit 3 generates the inversion data, which is the data read by the data reading unit 2 and which is divided into a certain time period, and which is inverted in the amplitude direction.

The level difference calculation unit 4 calculates the level difference (amplitude difference) from the data immediately before attempting to join the data divided into fixed time periods. The level difference calculation unit (1) 41 And a level difference calculation unit (2) 42.

The level difference calculation unit (1) 41 is for calculating the level difference between the data read by the data reading unit 2 and divided into fixed time periods, and the data immediately before joining.

The level difference calculation unit (2) 42 calculates the level difference between the inverted data and the data immediately before joining. The level difference determination unit 5 has a fixed time period in which the smaller data of the level differences calculated by the level difference calculation unit (1) 41 and the level difference calculation unit (2) 42 is read by the data reading unit 2. It is determined to be either the data delimited by the above or the inverted data obtained by inverting the data. The data with the smaller level difference is joined to the immediately preceding data, output as audio data (after conversion) 6, and stored.

As described above, the data obtained by dividing the digital audio data at regular time intervals is read, and the level difference between this data and the data immediately before attempting to join the data and the level difference immediately before attempting to join the inverted data obtained by inverting this data. The level difference from the data is compared, and the smaller data is joined to the immediately preceding data. This makes it possible to reduce discontinuity at the junction and reduce the noise level by joining the data immediately before joining and the data having a small level difference or the inverted data.

FIG. 1B shows a data joining method.
Here, the horizontal axis represents time and the vertical axis represents amplitude. When joining the data of the time of Ti to the immediately preceding data T (i-1), the same data as the source of the time of Ti, the inversion data inverted in the amplitude direction, and the data immediately before the attempt to join (time The level difference with the data of T (i-1) is as follows: -The difference between the same data as the original and the immediately preceding data is A (Q and P
Level difference A) The difference between the inverted data and the immediately preceding data is B (level difference B between R and P). Here, it is found that A> B, and the inverted data having a small level difference is joined to the immediately preceding data.

As described above, the data immediately before and the data having the same level as the original data or the inverted data, which has a small level difference, are repeatedly joined to perform the data joining, thereby reducing the discontinuity of the joined portion of the data. Therefore, the noise level can be reduced.

Next, the joining of the audio data in the configuration of FIG. 1 will be described in detail according to the order shown in the flowchart of FIG. FIG. 2 shows an overall control flowchart of the present invention.

In FIG. 2, S1 samples analog voice data and stores the digital value in a memory. In S2, the reproduction speed is designated. A value larger than 1 is specified for fast, and a value smaller than 1 is specified for slow.

In step S3, it is determined whether the reproduction speed is fast (> 1) or slow (<1). If fast (> 1), S4
Then, speed conversion processing (fast) is executed (FIG. 3), digital audio data is taken out one by one or plural times and joined to perform a reproduction speed. On the other hand, if it is slow (<1), the speed conversion process (slow) is executed in S5 (FIG. 4),
Digital audio data is spliced once or multiple times to slow down the playback speed.

In step S6, the voice data after joining is output.
As described above, in response to designation of a high reproduction speed (> 1) or a low reproduction speed (<1), audio data divided into digital constant time periods are extracted one by one or a plurality of times. Data with little level difference or inverted data is spliced to increase the playback speed, and one or more same data with audio data are repeated, and at this time, data with little level difference or inverted data is spliced and played. Slow down or
As a result, it is possible to reduce the level difference at the joint and reduce the noise level. The details will be sequentially described below.

FIG. 3 shows a speed conversion processing (acceleration) flowchart of the present invention. Here, the register represents the following. Register A: a register that stores the last value of output data (register that stores audio data immediately before attempting to join) Register B: a register that stores the value of the read data.

Register C: A register for storing inverted data inverted in the amplitude direction. In FIG. 3, S11
Initializes register A to 0. In S12, one block of digital audio data is read into the buffer.

In step S13, it is determined whether the data has ended. Yes
In the case of (when the reading of S12 fails due to the end of data), the process ends. On the other hand, if NO, S
Proceed to 14.

In S14, it is determined whether the data is skipped. If YES, the data is skipped,
Since it is not the data to be joined, the process returns to S12. NO
In the case of, it progresses to S15.

In step S15, the value at the head of the buffer is stored in the register B. S16 is the value of register A and register B
Determine if the values of are equal. This is because the last value of the output data stored in the register A (the value of the data immediately before trying to join) is equal to the first value read in the register B (= the level difference is zero). Determine. YE
In the case of S, the data read in S21 is output, and S
At 22, the last value of the output data is saved in register A and S
Return to 12. On the other hand, if NO in S16, the process proceeds to S17.

In S17, since the value stored in the read register B is not equal to the value stored in the register A in NO in S16, the value of the register B is inverted in the amplitude direction and stored in the register C. .

In S18, the value of register A and the value of register C
It is determined whether the value of is equal to. This is because the last value of the output data stored in the register A (the value of the data immediately before trying to join) is equal to the first value of the inverted data stored in the register C (= the level difference is zero). Determine whether.
In the case of YES, the inverted data is output in S21 and S2 is output.
The last value of the output data is saved in register A in 2 and S1
Return to 2. On the other hand, if NO in S18, the process proceeds to S19.

In step S19, it is determined whether or not | C-A |> | B-A |. This is because the absolute value | C-A | of the level difference between (reversed data) and (value of data immediately before attempting to join) is
It is determined whether or not it is larger than the absolute value | B-A | of the level difference between (the value of the first data read) and (the value of the data immediately before joining). In the case of YES, the read data having a small level difference is output in S21, the last value of the output data is stored in the register A in S22, and the process returns to S12. On the other hand, in the case of NO, inverted data having a small level difference is output in S20, the last value of the output data is stored in the register A in S22, and the process returns to S12.

As described above, the first value of the read data is stored in the register B, the first value of the inverted data is stored in the register C, and the last value of the previous output data is stored in the register A. However, these are compared, and when the value of the register A is equal to the value of the register B or | C−A |> | B−A |, the read data are joined.

When the value of the register A is equal to the value of the register C, or when | C−A | <| B−A |, the inverted data is joined. With these, by selecting one of the read data or inverted data that has the smallest level difference at the time of joining, and joining it, the level difference at the joint can be made smaller, the noise level can be reduced, and the speed can be increased. Is possible.

FIG. 4 shows a flow chart of the speed conversion process (slow down) of the present invention. Here, the register represents the following. Register A: a register that stores the last value of output data (register that stores audio data immediately before attempting to join) Register B: a register that stores the value of the read data.

Register C: A register for storing inverted data inverted in the amplitude direction. In FIG. 4, S31
Initializes register A to 0. In S32, one block of digital audio data is read into the buffer.

In step S33, it is determined whether the data has ended. Yes
In the case of (when the reading of S32 fails due to the end of the data), the processing ends. On the other hand, if NO, S
Proceed to 34.

In S34, the value at the head of the buffer is stored in the register B. S35 is the value of register A and register B
Determine if the values of are equal. This is because the last value of the output data stored in the register A (the value of the data immediately before trying to join) is equal to the first value read in the register B (= the level difference is zero). Determine. YE
In the case of S, the data read in S40 is output, and S
The last value of the output data is stored in register A at 41, and S
Proceed to 42. On the other hand, in the case of NO in S35, the process proceeds to S36.

In S36, since the value stored in the read register B and the value in the register A are not equal to each other in NO in S35, the value in the register B is inverted in the amplitude direction and stored in the register C. .

In S37, the value of register A and the value of register C
It is determined whether the value of is equal to. This is because the last value of the output data stored in the register A (the value of the data immediately before trying to join) is equal to the first value of the inverted data stored in the register C (= the level difference is zero). Determine whether.
If YES, the buffer data is inverted in the amplitude direction in S39, the block data is output in S40, and S4 is output.
The last value of the output data is saved in register A by 1 and S4
Go to 2. On the other hand, if NO in S37, the process proceeds to S38.

In step S38, it is determined whether or not | C-A |> | B-A |. This is because the absolute value | C-A | of the level difference between (reversed data) and (value of data immediately before attempting to join) is
It is determined whether or not it is larger than the absolute value | B-A | of the level difference between (the value of the first data read) and (the value of the data immediately before joining). If YES, the block data is output in S40, the last value of the output data is stored in the register A in S41, and the process proceeds to S42. On the other hand, in the case of NO, the data in the buffer is inverted in the amplitude direction in S39, the block data is output in S40, the last value of the output data is stored in the register A in S41, and the process proceeds to S42.

A step S42 decides whether or not the output of the block is completed. This determines whether the output of the buffer is completed according to the specified reproduction speed, for example, whether one buffer is output twice when half the reproduction speed of the original data is specified. . In the case of YES, the process is finished, and the process returns to S32. In the case of NO, since it has not ended, the process returns to S34.

As described above, the first value of the read data is stored in the register B, the first value of the inverted data is stored in the register C, and the last value of the previous output data is stored in the register A. However, these are compared, and when the value of the register A is equal to the value of the register B, or when | CA->> | BA-, the read data is joined a specified number of times.

When the value of the register A is equal to the value of the register C, or when | C−A | <| B−A |, the inverted data is joined a specified number of times. With these, the read data or the inverted data, whichever has the smallest level difference when joined, is selected and joined a specified number of times to reduce the level difference at the joint and reduce the noise level. It is possible to slow down the speed.

FIG. 5 shows an example of reproducing speed conversion (for increasing speed) of the present invention. This playback speed conversion example sets the playback speed to 2
An example of double conversion is shown below. The horizontal axis represents time and the vertical axis represents amplitude.

FIG. 5A shows original data (original data). FIG. 5B shows an example in which the reproduction speed is doubled. In this case, every other original data shown in FIG. 5A, the original data or the inverted data of this original data, which has a small level difference from the immediately preceding data, is selected. here,
The part described as inversion is a combination of inversion data.

As described above, when the reproduction speed is increased, the original digital data is divided into a certain time period, and when the divided data is joined every other here, the level difference from the immediately preceding data is generated. Either the small original data as it is or the inverted data that has been inverted is selected and joined. As a result, the frequency of discontinuity at the time of joining and the level of the discontinuity are lowered, and the noise level can be reduced.

FIG. 6 shows an example of reproduction speed conversion of the present invention (when it is slowed down). This playback speed conversion example sets the playback speed to 1
An example of conversion to / 2 is shown. The horizontal axis represents time and the vertical axis represents amplitude.

FIG. 5A shows original data (original data). FIG. 5B shows an example in which the reproduction speed is reduced to 1/2. This is one in which the original data of FIG. 6A or the inverted data of this original data, which has a small level difference from the immediately preceding data, is selected and joined twice.

As described above, when the reproduction speed is slowed down, the original digital data is divided into a certain time period, and when the divided data are joined, the original data having a small level difference from the immediately preceding data is kept as it is. Alternatively, one of the inverted data which has been inverted is selected and joined twice, for example. As a result, the frequency of discontinuity at the time of joining and the level of the discontinuity are lowered, and the noise level can be reduced.

[0054]

As described above, according to the present invention,
When synthesizing audio data that has been segmented into fixed time periods, the data that has the smaller level difference between the immediately preceding data and the segmented or inverted data can be spliced to speed up or slow down the playback speed. By adopting the configuration described above, it is possible to reduce the frequency of occurrence of discontinuous portions of the joint portion, reduce the level of the discontinuous portions, and reduce noise.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an embodiment of the present invention.

FIG. 2 is an overall control flowchart of the present invention.

FIG. 3 is a flow chart of speed conversion processing (speeding up) of the present invention.

FIG. 4 is a speed conversion processing (slowing) flowchart of the present invention.

FIG. 5 is an example of reproduction speed conversion (when increasing the speed) of the present invention.

FIG. 6 is an example of reproduction speed conversion (when slowing down) according to the present invention.

FIG. 7 is an example of a conventional reproduction speed conversion (when increasing the speed).

FIG. 8 is an example of a conventional reproduction speed conversion (when it is slowed down).

[Explanation of symbols]

 1, 6: Audio data 2: Data reading unit 3: Data inversion processing unit 4, 41, 42: Level difference calculation unit 5: Level difference determination unit

Claims (3)

[Claims] 1. A reproduction speed conversion method for dividing audio data into fixed time periods, joining the data, and converting the reproduction speed, wherein the audio data is divided into constant time periods, and inverted data obtained by inverting the data. A reproduction speed conversion method, characterized in that the reproduction speed of audio data is converted by repeating the calculation of the level difference from the data to be combined and combining the data with the smaller level difference. 2. The reproduction speed conversion according to claim 1, wherein the data obtained by dividing the audio data into a predetermined time period is joined one by one or a plurality of times to increase the reproduction speed. Method. 3. The reproduction according to claim 1, characterized in that, with respect to data obtained by dividing the audio data into a certain time period, the same data is joined once or a plurality of times to reduce the reproduction speed. Speed conversion method.