JP3035948B2 - Audio data playback method - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to an audio data reproducing method for reproducing an audio signal converted into PCM data by quasi-instantaneous compression.

[Summary of the Invention]

The present invention relates to an audio data reproducing method for reproducing an audio signal converted into PCM data by quasi-instantaneous compression. The PCM data is in a mode irrelevant to the audio data, and is reproduced from the timing at which PCM data in a mode irrelevant to the immediately preceding audio data is obtained, so that good audio is always reproduced.

[Conventional technology]

Conventionally, when converting an audio signal into a digital signal as PCM data, in order to efficiently convert the data, for example, Japanese Patent Application No. Sho 63
As described in -301544, conversion to PCM data by quasi-instantaneous compression has been proposed. That is, as shown in FIG. 5, for example, the data compression ratio of the block-formed PCM data is changed for each block, and additional information (compression-related parameters) is added in addition to the compressed data. And the filter on the decoder side (reproduction side) is II
An R (cyclic) digital filter configuration is obtained.

Here, illustrating the configuration of each block in FIG. 6, 1 as the block of data, 1-byte header information (parameter information) RF and 8 bytes of sample data D _A0 to D _B3
It is composed of The header information RF includes 4-bit range information, 2-bit filter selection information, and two 1-bit flag information, for example, information indicating a block including a loop start point (loop start flag LSF). ) And information (loop end flag LEF) indicating the block at the end point of the loop.
Here, the peak value data of one sample is bit-compressed and represented by 4 bits, and the data D _{A0 to} D _B3 include 4-bit data D _{A0H to} D _{B3L for} 16 samples. .

Then, the filter on the decoder side is selected from the 0th-order filter, the 1st-order filter, and the 2nd-order filter according to the 2-bit filter selection information. That is, the repetition number of the IIR filter on the reproduction side is selected from 0 to 2 based on the filter selection information. In this case, when the number of tours is 0, the so-called straight PCM mode is performed in which no tour is performed and there is no correlation with the data of the previous block.

[Problems to be solved by the invention]

By the way, when the data thus compressed is reproduced from the beginning, the data is restored to the original number of bits by processing such as cycling by an IIR filter, and a good sound is reproduced. When the reproduction is started from the beginning, the predetermined tour set in the data at the time of starting the reproduction is not performed, and there is a possibility that an unnatural sound may be reproduced.
That is, when the data is circulated by the IIR filter, the data of the next block is added to the circulated data of the predetermined block, and the original data is obtained. However, when played back from the middle, the original data is not restored in this way,
Noise that cannot be heard is reproduced.

It is an object of the present invention to reproduce good sound when such quasi-instantaneously compressed data is reproduced from an arbitrary position.

[Means for solving the problem]

The audio data reproducing method of the present invention uses a quasi-instantaneous compression
In the audio data reproducing method of reproducing a continuous audio signal converted into M data, the audio signal at the part where the human voice is recorded and the part at the break of the phrase are forcibly changed to have no relation to the immediately preceding audio data. Mode PCM data,
When playing this PCM data from any point, muting is performed until PCM data in a mode unrelated to the previous audio data is obtained, and PCM data in a mode unrelated to the previous audio data is obtained It is made to play from.

In addition, an audio signal other than a human voice is always set as PCM data in a mode different from the immediately preceding audio data on a channel different from the audio signal of the human voice, and when performing the above-described reproduction, this PCM data is used. It is designed to be mixed with PCM data of human voice converted to PCM data by quasi-instantaneous compression and reproduced.

Furthermore, a plurality of channels of PCM data converted to PCM data by quasi-instantaneous compression including a channel in which a human voice is recorded.
When playing CM data from an arbitrary location at the same time, the channel where the human voice is recorded is played from the timing when the PCM data in the mode irrelevant to the previous audio data is played back, and the PCM data of other channels is At the same timing, the reproduction is forcibly regarded as a mode unrelated to the immediately preceding audio data.

[Action]

By doing this, when random playback is performed from an arbitrary point, playback is performed from the timing at which PCM data in a mode irrelevant to the previous audio data is played,
The sound data related to the immediately preceding sound data is not reproduced from the middle and unnatural sound is not reproduced, and good sound is always reproduced.

〔Example〕

An embodiment of the audio data reproducing method according to the present invention will be described below with reference to FIGS.

First, FIG. 2 shows an example of an analog tone signal waveform before sampling. As is clear from the analog waveform of FIG. 2, immediately after the start of the sounding of a general musical instrument sound, the periodicity of the waveform is reduced due to the inclusion of non-pitch components such as a keystroke noise of a piano and a breath noise of a wind instrument. A formant portion FR, which is an unclear portion, often occurs, and thereafter, the same waveform repeatedly appears in a basic cycle corresponding to a musical interval (pitch, pitch). Predetermined n cycles of the repeating waveform (n is an integer) a a looping section LP, represents this section LP start point and looping start point the end point, respectively LP _S and Rubingu end point LP _E (looping points). Then, the digital data corresponding to the analog waveform of the formant part FR and the looping section LP is recorded on a storage medium, and at the time of reproduction, the looping section LP is repeatedly reproduced following the reproduction of the formant part FR, for an arbitrary long time. Can generate music.

Here, the music data including such a looping section LP is bit-compressed and encoded prior to being stored in a storage medium such as a memory, so that the data amount is reduced. The bit compression encoding method proposed by the present applicant in Japanese Patent Application Laid-Open Nos. 61-158217, 62-8629 or 62-3516, that is, a method of determining the peak value data. It is assumed that a high-efficiency encoding method is used in which blocks are formed for every number of samples (h samples) and optimal bit compression is performed in units of these blocks.
This will be schematically described with reference to FIG.

In FIG. 1, the high-efficiency bit compression encoding system includes an encoder (10) on the recording side and a decoder (30) on the reproducing side, and an input terminal (11) of the encoder (10) is , The peak value data x (n) of the tone signal.

This input signal (peak value data) x (n) is supplied to an FIR (finite impulse response) digital filter (14) composed of a predictor (12) and an adder (13).
Predicted signal from the predictor (12) (peak value data)
(N) is sent to the adder (13) as a subtraction signal.
The adder (13) subtracts the prediction signal (n) from the input signal x (n) to output a prediction error signal or a difference output d (n) in a broad sense. The predictor (12) generally has p past inputs x (n-p), x (n
−p + 1), ‥‥, x (n−1) to calculate the predicted value (n). Note that the FIR filter (1
4) is hereinafter referred to as an encoding filter.

In this bit compression encoding system, an input signal (peak value data) x (n) is divided into blocks for a predetermined number of samples (h samples), and an encoding filter (14) having optimum characteristics for each block. To choose. In this method, a plurality of (for example, four) encoding filters having different characteristics are provided in advance, and a filter having an optimum characteristic, that is, a filter capable of obtaining the highest compression ratio is selected from these filters. It can be realized by doing. However, general digital
In terms of the configuration of the filter, one encoding unit shown in FIG.
A plurality of sets (for example, four sets) of coefficients of the predictor (12) of the filter (14) are stored in a coefficient memory or the like, and these sets of coefficients are switched and selected in a time-sharing manner. In many cases, an operation equivalent to selecting one of a plurality of encoding filters is performed. As a specific example,
A total of four types of encoding filters, ie, 0 order, 1 order, and 2 types of secondary, are prepared, and 2-bit filter selection information (mode selection information) indicating which of these 4 types of filters has been selected is provided. What is necessary is just to transmit by block. Selecting a zero-order encoding filter corresponds to selecting a straight PCM mode in which an input PCM signal is output as it is.

Next, a difference output d (n) as a prediction error is passed through an adder (21) to a gain G shifter (15) and a quantizer (16).
Compression processing or ranging in which, for example, the exponent part in the floating point display form corresponds to the gain G, and the mantissa part corresponds to the output from the quantizer (16). Processing is performed. That is, the range is switched by shifting the input data of, for example, 16 bits by the shifter (16) by the number of bits according to the gain G, and the fixed number of bits of the data bit-shifted by the quantizer (16), for example, 4 Requantization is performed to extract bits. Here, the noise shaping circuit (noise shaper) (17)
Obtains a so-called quantization error by an adder (18) corresponding to an error between an output and an input of the quantizer (16), and obtains the quantization error by a gain.
The signal is sent to the predictor (20) via the G- ¹ shifter (19), and so-called error feedback is performed in which the prediction signal of the quantization error is fed back to the adder (21) as a subtraction signal. In this way, requantization by the quantizer (16) and noise
The error feedback is performed by the shaping circuit (17), and the output (n) is taken out from the output terminal (22).

The output d '(n) from the adder (21) is obtained by subtracting the prediction signal (n) of the quantization error from the noise shaper (17) from the difference output d (n).
The output d "(n) from the shifter (15) is obtained by multiplying the gain G by the output d '(n) from the output adder (21). The output from the quantizer (16) ( n) is the sum of the quantum error e (n) in the quantization process and the output d ″ (n) from the shifter (15), and the quantization error in the adder (18) of the noise shaper (17). e (n)
Is taken out. This quantization error e (n) is calculated by the gain G ^-1
And a prediction signal (n) of a quantization error by passing through a predictor (20) that takes a linear combination of the past r inputs through the shifter (19).

The sound source data is subjected to the above-described encoding processing, becomes an output (n) from the quantizer (16), and is taken out via the output terminal (22).

Next, from the prediction / range adaptation circuit (24), mode selection information as optimal filter selection information is output and sent to, for example, a predictor (12) and an output terminal (27) of the encoding filter (14). Further, the gain G and the gain G- ¹ or the range information for determining the bit shift amount are output and sent to the shifters (15) and (19) and the output terminal (26). These mode (filter) selection information and range information are also referred to as parameters relating to compression.

Next, the output (n) from the output terminal (22) of the encoder (10) is transmitted to the input terminal (31) of the decoder (30) on the reproduction side, or obtained by being recorded and reproduced. Signal '(n) is provided. This input signal '(n) is sent to the adder (33) via the shifter (32) having a gain of G- ¹ . The output x '(n) from the adder (33) is sent to the predictor (34) and
(N), and the predicted signal '(n) is added to the adder (33).
To be added to the output "(n) from the shifter (32). The added output is output from the output terminal (35) as a decoded output '(n).

The range information and mode selection signal output from the output terminals (26) and (27) of the encoder (10) and transmitted, recorded, or reproduced are input to the input terminals (36) and (37) of the decoder (30). ). The range information from the input terminal (36) is sent to the shifter (32) to determine the gain G- ^1, and the mode selection information from the input terminal (37) is sent to the predictor (34) to specify the prediction. To determine. The prediction characteristic of the predictor (34) is selected to be equal to the characteristic of the predictor (12) of the encoder (10).

In the decoder (30), the output "(n) from the shifter (32) is obtained by multiplying the input signal '(n) by the gain G ^-1 . The output' (n) of the adder (33) )
Is the output "(n) from the shifter (32) and the prediction signal '
(N) is added.

In such a configuration bit compression coding system, for example, at the time of recording of the instrument sound as shown in FIG. 2, the first predetermined block (looping start block) starting from the looping start point LP _S looping section LP of FIG. 2 Let the number words be words of straight PCM data. This predetermined number may be equal to or greater than the maximum value of the order of the encoding filter (14) (order of the prediction filter). More specifically, the maximum order of the encoding filter (14) is quadratic. Therefore, it is sufficient to arrange the words W of two or more straight PCM data. For this purpose, for example, a changeover switch is provided immediately before the encode output terminal (22) so that the compressed data from the quantizer (16) and the straight PCM data can be selected and output, and the looping start point LP _S It is sufficient to select and output straight PCM data by a changeover switch during a predetermined number of words from. Since the range information is independent for each block, instead of outputting the 16-bit straight PCM data having the word length of the original sample as it is, for example, a 4-bit straight data obtained by compressing only the bits corresponding to the range information of the same block is used. It can also output PCM data.

In this example, when the conversation data is recorded (stored) as voice data, the beginning part of each conversation sentence and the break between phrases are set as straight PCM data. In other words, it is, for example, "Hello, good weather
In the case of conversation data to be pronounced goodbye ", as shown in FIG. 3, which is the first part of each statement and the top part a of" Hello ", and the top portion (b) of the" sounds good weather ", of" goodbye " The head c is recorded as straight PCM data, and as a decode output supplied to the terminal (35) of the decoder (30) shown in FIG.
It is assumed that the audio data has not been looped by 4) and the adder (33).

When reproducing such conversation data, when performing so-called random reproduction in which reproduction is performed from an arbitrary position, reproduction is always performed from a position recorded as straight PCM data. That is, as shown in the flowchart of FIG. 4, when audio data is decoded at the time of reproduction, whether the decoding at this time is reproduction from an arbitrary position by random reproduction or continuous reproduction from previous data is determined. It is determined whether the playback operation is continuous and the playback operation is continued. When random reproduction is performed, a mode selection signal in the reproduced audio data is determined, and muting is performed to inhibit output of the reproduced audio until a mode selection signal of straight PCM data is detected. When the mode selection signal of the straight PCM data is detected, the mute is released and the reproducing operation is started.

Since random playback is performed in this manner, the point at which random playback is started is the timing at which PCM data in a mode irrelevant to the immediately preceding audio data is obtained, so that the data is cyclically output. The PCM data is not output without being circulated, and there is no possibility that the output sound at the start of the reproduction will be disturbed. In actual conversation data, since the beginning of a conversation sentence or a break in a phrase frequently appears, muting is often performed in a short time.

Here, an actual use state of an audio signal converted into PCM data by such quasi-instantaneous compression will be described.For example, an audio signal recorded together with a video signal on a video disc on which a movie is recorded is compressed as described above. PCM data. In this case, a plurality of channels are prepared as audio signals, and conversations of characters are recorded as PCM data of different channels in a plurality of languages. The background sound is recorded as PCM data on another channel. At the time of playback, one of the prepared languages is selected, and the PCM of the language of the selected channel is selected.
The data and PCM data of the background sound channel are decoded, and the audio signals of both channels are mixed and reproduced. By setting a plurality of channels using the PCM data compressed in this way, the amount of audio data can be significantly reduced, and recording on a video disc for a plurality of channels becomes possible.

Then, by applying the above-described audio data reproducing method of the present example to a device that performs recording / reproduction in this manner, at the time of random reproduction, reproduction is always performed from the straight PCM mode, and the reproduction voice of the conversation at the start of reproduction is reproduced. There is no disturbance. In this case, the background sound PCM data is always straight PCM
By recording in the mode, the background sound can be reproduced well.

Also, if the PCM data of the background sound is also converted into PCM data by semi-instantaneous compression and recorded, the channel of the conversation data at the time of random playback is played from the beginning of the conversation sentence or the break of the phrase in which the PCM mode is straight PCM mode. Even in the mode in which the PCM data of the background sound circulates, the playback may be forcibly regarded as the straight PCM mode. In this case, the first part of the background sound is slightly distorted, but unlike the conversational sentence, there is no need to listen to the contents, so there is almost no problem.

Note that the present invention is not limited to the above-described embodiment, but may take various other configurations.

〔The invention's effect〕

According to the present invention, the speech signal of the part where the conversation of the human voice is recorded and the break of the phrase is set as PCM data in a so-called straight PCM mode irrelevant to the immediately preceding audio data, and random reproduction is performed from an arbitrary position. When performing, since it was made to play from the timing that became the PCM data of this straight PCM mode, audio data related to the immediately preceding audio data is played from the middle and unnatural audio is not played , Always good sound is played.
In this case, even when randomly reproducing the PCM data of a plurality of channels including the channel where the human voice was recorded from an arbitrary position at the same time, the channel where the human voice was recorded became the PCM data in the straight PCM mode. Since the reproduction is performed from the timing, a good sound is always reproduced.

[Brief description of the drawings]

FIG. 1 is a block diagram of a system based on an audio data reproducing method according to one embodiment of the present invention, FIGS. 2 and 3 are explanatory diagrams showing data examples of one embodiment, and FIG. FIGS. 5 and 6 are flowcharts for explanation, and are explanatory diagrams showing examples of audio data. (30) is a decoder, (33) is an adder, and (34) is a predictor.

Continuation of front page (56) References JP-A-62-3516 (JP, A) JP-A-60-113300 (JP, A) JP-A-58-215697 (JP, A) JP-A-3-265309 (JP) , A) Japanese Patent No. 2674161 (JP, B2) Japanese Patent Publication No. 6-1903 (JP, B2) Japanese Patent Publication No. 6-101709 (JP, B2) Japanese Patent Publication No. 59-12187 (JP, B2) Japanese Patent Publication No. 6-44200 (JP, B2) JP, B2) JP 8-21200 (JP, B2) (58) Fields investigated (Int. Cl. ⁷ , DB name) G10L 19/00 G10H 7/02 G11B 20/10 301 H03M 1/16 H04B 14 / 04 INSPEC (DIALOG) JICST file (JOIS) WPI (DIALOG)

Claims (3)

(57) [Claims] An audio data reproducing method for reproducing a continuous audio signal converted into PCM data by quasi-instantaneous compression, wherein a speech at a place where a human voice is recorded and an audio signal at a break between phrases are forcibly transmitted. The PCM data in a mode irrelevant to the immediately preceding audio data is used.When this PCM data is reproduced from an arbitrary position, muting is performed until PCM data in a mode irrelevant to the immediately preceding audio data is obtained. An audio data reproduction method that reproduces from a timing at which PCM data in a mode irrelevant to the audio data is obtained. 2. An audio signal other than a human voice is converted into PCM data in a mode different from that of the immediately preceding audio data on a channel different from that of the human voice audio signal. 2. The audio data reproducing method according to claim 1, wherein the audio data is reproduced by being mixed with the digitized human voice PCM data. 3. A multi-channel PCM converted into PCM data by quasi-instantaneous compression including a channel in which a human voice is recorded.
When playing data from an arbitrary location at the same time, the channel on which the human voice was recorded becomes the PCM data in a mode irrelevant to the previous audio data, and the PCM data of other channels are played back at the same time. 2. The audio data reproducing method according to claim 1, wherein the reproduction is forcibly regarded as a mode unrelated to the immediately preceding audio data from the timing.