JPH09146593A - Methods and devices for sound signal coding and decoding - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make it possible to efficiently code sound signals even at a low bit rate by sub-band-coding sound signals and extracting features by short-time frames from the obtained narrow band signals. SOLUTION: Inputted PCM audio signals are supplied also to a sub-band filter 11 in parallel with a FET circuit 13. In the sub-band filter 11, the signals are equally divided into, for example, 32 sub-bands and, in addition, supplied to a frame processing circuit 12 for every sub-band to be divided into frame of every specified period of time and outputted to an envelope extracting circuit 19. This circuit 19 extracts envelopes of amplitude calculated for every short- time frame. And, using a zero-cross counting circuit 12, the number of zero crossings of amplitude is calculated for every short-time frame. Thus, the envelopes quantized for every sub-band and the number of zero-crossings are supplied to a multiplexer 16 to be multiplexed and outputted to the transmission line as bit stream.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a signal processing device for compressing the information amount of a digital (PCM) signal, and more particularly to a compression encoding / decoding device for digital audio signal.

[0002]

2. Description of the Related Art Conventionally, subband coding and orthogonal transform coding, which are mainly used for high-efficiency coding of audio signals, map a sample signal in the time domain to the frequency domain and calculate the energy of the frequency domain. It is a method of encoding while controlling the bit allocation amount due to uneven distribution by utilizing human psychoacoustic characteristics.

A typical MPEG in such an encoding method
An example of an audio signal encoding device and audio decoding device of Audio is shown in FIGS. 5 (A) and 5 (B). The audio signal input to the audio signal encoding device shown in FIG. 5A is equally divided into 32 bands by the band division filter 1. Then, for each signal divided into 32 bands by the band division filter 1, a maximum value (scale factor) is obtained for each analysis frame and normalized (scaling circuit 2).

Separately from this, the input audio signal is orthogonally transformed by the FFT circuit 3, masking is calculated based on the psychoacoustic model (masking calculation circuit 4), and the bit amount of 32 bands is determined. Then, the determined bit amount is quantized by a quantizer 5 provided for each frame of 32 bands (frame data), and multiplexed with a scale factor (multiplexer circuit) 6 by a multiplexer (bit multiplexing circuit) 6 to form a bit stream. Is transmitted.

On the side of the audio signal decoding apparatus shown in FIG. 5B, the supplied bit stream is decomposed into a scale factor and frame data by a demultiplexer (bit decomposition circuit) 7, and the frame data (band signal) is decomposed. An inverse quantizer 8 performs inverse quantization, an inverse scaling circuit 9 performs scale adjustment for each band by a scale factor, and a band synthesis filter 10 synthesizes signals in all bands to output an audio signal. To be done.

[0006]

The audio signal coding apparatus as described above is particularly applied to coding of high quality audio, and realizes good quality at a bit rate of 128 kbps or higher. However, at a low bit rate of 64 kbps or less, which is required for communication applications and the like, the coding quality is deteriorated, so a more efficient method of analyzing and describing an acoustic signal has been desired.

The present invention has been made in view of the above-mentioned problems of the prior art. The subband coding method is further provided with a means for analyzing a band signal to extract a characteristic and extract it. An object of the present invention is to provide an encoding method and an encoding apparatus for encoding an audio signal with higher efficiency than before by encoding, and a decoding method and a decoding apparatus for decoding the same.

[0008]

In a subband coding method such as MPEG Audio, the code amount is reduced by allocating bits to a band-divided signal by psychoacoustic analysis, and the time of sampling the band signal is reduced. Redundancy is not considered. Therefore, by taking the temporal redundancy of this band signal into consideration, it is possible to perform more efficient encoding than the conventional subband encoding method.

The acoustic signal coding apparatus of the present invention divides the supplied acoustic signal into bands by a subband coding method,
Features are extracted from the obtained narrowband signal in a short time frame and encoded. Then, the subband-converted narrowband signal is encoded by the following procedures (1) to (3). (1) An envelope of amplitude is extracted for each frame for a short time. (2) The number of times the amplitude waveform crosses zero in the frame,
Or measure the time interval of zero crossing. (3) The contents of (1) and (2) above are encoded and transmitted as a feature of the frame.

In the acoustic signal decoding device, the subband narrowband signal is analyzed in a short time frame, and if the number of zero crossings is transmitted in (2) above, a sine wave is generated for each frame in each band. Is generated, shaped by the envelope transmitted in (1) above, and decoded. When the zero-crossing time interval is transmitted in (2) above, the zero-crossing point and the point on the envelope of the time between the zero-crossing points transmitted in (1) above are interpolated with an appropriate curve. And decrypt.

At this time, since the envelope information obtained in (1) above is considered to have a correlation in each band, it is possible to reduce the amount of information by taking a difference. Furthermore, the amount of information can be significantly reduced by vector-quantizing the outline of the envelope in a short time. In order to improve the coding quality, the change in the time interval is also transmitted when transmitting not the number of zero crossings in the above (2) but the change in the time (sample) interval of the zero crossings in time series. By vector-quantizing the time series, it is possible to improve the coding quality without significantly increasing the code amount.

[0012]

BEST MODE FOR CARRYING OUT THE INVENTION The acoustic signal coding method of the present invention will be described with reference to the acoustic signal graph shown in FIG. The supplied acoustic signal is as shown in the graph of FIG.
For example, it is band-divided into bands 1 to 4, and the band-divided narrowband signals are respectively cut out into short-time analysis frames 1 to 3. Then, the envelope (dashed-dotted line in the figure) of each narrowband signal is extracted by the envelope extraction circuit for each frame. Then, the number of zero crossings of the amplitude or the zero crossing interval is calculated for each frame of each band. At this time, when the zero crossing occurs between the frames, the number of zero crossings is counted only in one frame. The envelope outline (one-dot chain line in the figure) obtained for each frame in each band and the number of zero crossings or the zero crossing interval are encoded as characteristic parameters of the frame.

Further, according to the audio signal decoding method of the present invention, when a coded number of zero crossings is transmitted as a characteristic parameter, after dequantization, a sine wave is generated in the restored frame to generate an envelope. A short-time frame is restored by shaping the sine wave with a line. If a coded zero crossing interval is transmitted as a feature parameter, after inverse quantization, the interpolation position on the envelope is calculated from the time interval of the zero crossing, and the zero crossing point and the interpolation on the envelope are interpolated. A short time frame is restored by interpolating the position with a smooth curve. After that, the acoustic signal is decoded by synthesizing each frame and each band.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Some embodiments of an audio signal coding apparatus and a decoding apparatus of the present invention will be described with reference to the drawings.

<Embodiment 1> FIGS. 1A and 1B are block diagrams showing a first embodiment of an audio signal encoding apparatus and a decoding apparatus according to the present invention, respectively. In the first embodiment, a configuration in which the number of zero crossings is used will be described. In the acoustic signal encoding device shown in FIG.
The M audio signal is supplied to the FFT circuit 13 and orthogonally transformed, and then supplied to the masking calculation circuit 14 and bit allocation (masking) based on a known psychoacoustic model (used in MPEG Audio and the like).
Is performed.

The input PCM audio signal is also supplied to the well-known band division filter 11 in parallel with the FFT circuit 13. The band division filter 11 is equally divided into, for example, 32 bands. Further, it is supplied to the frame processing circuit 12 for each of the bands, divided into frames for every fixed time (for example, every 256 samples or every 1468 samples), and output to the envelope extraction circuit 19. The envelope extraction circuit 19 calculates and extracts the envelope of the amplitude for each short time frame as shown in FIG. 4 by the above-described acoustic signal encoding method. Further, the zero-crossing counting circuit 21 calculates the number of zero-crossings of the amplitude for each short-time frame.

The bit allocation output of the masking calculation circuit 14, the envelope for each short time frame output from the envelope extraction circuit 19, and the number of zero crossings for each short time frame output from the zero crossing counting circuit 21. And is the quantizer 1
5, and the envelope is quantized with the extraction accuracy based on the bit allocation output supplied from the masking calculation circuit 14. Further, the envelope and the number of zero crossings thus quantized for each band are supplied to the multiplexer 16 to be multiplexed and output to the transmission line as a bit stream.

In the acoustic signal decoding apparatus shown in FIG. 1B, the bit stream sent from the transmission line is separated by the demultiplexer 17 into the number of zero crossings and the envelope curve for each frame in each band, and the inverse quanta is calculated. It is supplied to the quantizer 18 and inversely quantized. After that, the number of zero crossings is supplied to the sine wave generation circuit 22 to generate a sine wave for each frame, and the envelope is supplied to the envelope restoration circuit 23 to be restored, and each is supplied to the mixing circuit 24. , By shaping the sine wave with the envelope, the short-time frame for each band is restored. The signal restored for each frame in each band in this manner is supplied to the band synthesizing filter 20, is synthesized into a full band signal, and is output as a PCM audio signal.

<Embodiment 2> FIG. 2 is a block diagram showing a second embodiment of the audio signal encoding apparatus of the present invention. In the second embodiment, the difference circuit 25 is added to the above-described first embodiment. The same components as those in the first embodiment are designated by the same reference numerals and the description thereof will be omitted. In the acoustic signal coding apparatus shown in the figure, a difference circuit 25 is provided between the envelope extraction circuit 19 and the quantizer 15 in the first embodiment of the acoustic signal coding apparatus shown in FIG. The difference circuit 25 includes an envelope from the envelope extracting circuit 19 and an envelope extracting circuit 19 of an adjacent subband (band).
And the envelope curve from (1) is supplied, and after taking the difference (residual) between these, the residual is output to the quantizer 15 and quantized. With this configuration, the transmission code amount can be further reduced. Note that the acoustic signal decoding device in this case may be configured so that the envelope restoration circuit 23 restores the envelope from the residual of the envelope.

<Embodiment 3> FIGS. 3A and 3B are block diagrams showing a third embodiment of the audio signal coding apparatus and the decoding apparatus of the present invention, respectively. In the third embodiment, a configuration using a zero crossing interval will be described. In the acoustic signal encoding device shown in FIG.
The M audio signal is supplied to a known band division filter 31. The band division filter 31 divides the band into, for example, 32 bands. Further, it is supplied to the frame processing circuit 32 for each band, divided into frames at constant time intervals (for example, every 33.3 msec which is an NTSC video frame), and output to the envelope extraction circuit 39. The envelope extraction circuit 39 calculates and extracts the envelope of the amplitude for each short time frame as shown in FIG. 4 by the acoustic signal encoding method described above. Further, the zero-crossing interval detection circuit 34 calculates the zero-crossing interval of the amplitude for each short-time frame.

The envelope output from the envelope extraction circuit 39 is supplied to the vector quantizer 35a, an envelope similar to this envelope is found from the codebook, and output as an index corresponding to this envelope. To be done. Similarly, the zero crossing interval is also set in another vector quantizer 35b.
And the index corresponding to this zero crossing interval is output. The envelopes output from the vector quantizers 35a and 35b and the indexes corresponding to the zero-crossing intervals are supplied to the multiplexer 36, multiplexed, and output to the transmission path as a bit stream.

In the acoustic signal decoding apparatus shown in FIG. 1B, the bit stream sent from the transmission line is separated by the demultiplexer 37 into the zero crossing intervals and the envelope index of each frame in each band. , The vector dequantizers 38a and 38b, respectively, and the same codebook as that of the encoder is used, and the zero crossing intervals and the envelopes corresponding to the respective indexes are restored to the zero crossing interval restoring circuit 42 and the envelope restoring. It is supplied to the circuit 43 and restored.

The zero crossing interval and the envelope output from the zero crossing interval restoring circuit 42 and the envelope restoring circuit 43 are
It is supplied to the interpolation point calculation circuit 41, the interpolation points on the envelope are calculated from the zero-crossing intervals, and the interpolation circuit 44 interpolates the zero-crossing points and the interpolation points on the envelope with a smooth curve. The short-time frame of is being restored. The signal thus restored for each frame in each band is supplied to the band synthesizing filter 40, synthesized into a full band signal, and output as a PCM audio signal.

As a modification of this embodiment, in the acoustic signal coding apparatus, a zero-crossing counting circuit for calculating the number of zero-crossings instead of detecting the zero-crossing intervals by the zero-crossing interval detection circuit 34 is used. May be encoded to reduce the code amount. Further, in the audio signal encoding device, the timing for cutting out a short time frame may be set independently in all bands,
Further, the timing for cutting out the frames of the adjacent bands may be set to be shifted or may be set to be synchronized.

[0025]

The acoustic signal coding apparatus of the present invention can improve the compression efficiency because the parameters to be transmitted can be described more easily than the conventional subband coding apparatus. Further, the acoustic signal encoding device of the present invention,
Even at a lower bit rate than the conventional sub-band encoder,
A sound signal of good quality can be transmitted.

When the difference between the envelope information of adjacent bands is calculated, the amount of information can be further reduced. Furthermore, the amount of information can be significantly reduced even when the envelope information is vector-quantized. Also, in order to improve the coding quality, even when transmitting a time series showing the change in the time (sample) interval of the zero crossings instead of the number of zero crossings, the time series of the change in this time interval is represented by a vector quantum. It is possible to improve the coding quality without significantly increasing the code amount.

The acoustic signal decoding device of the present invention is
There is an effect that a signal encoded by the acoustic signal encoding device described above can be decoded to obtain an acoustic signal of good quality.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing a first embodiment of an acoustic signal encoding device and a decoding device of the present invention.

FIG. 2 is a configuration diagram showing a second embodiment of the acoustic signal encoding device of the present invention.

FIG. 3 is a configuration diagram showing a third embodiment of the audio signal encoding device and the audio signal encoding device of the present invention.

FIG. 4 is a graph for explaining an acoustic signal coding method and a decoding method of the present invention.

FIG. 5 is a configuration diagram showing a conventional acoustic signal encoding device and decoding device.

[Explanation of symbols]

 1, 11, 31 Band division filter 2 Scaling circuit 3,13 FFT circuit 4,14 Masking calculation circuit 5,15 Quantizer (encoding means) 6,16,36 Multiplexer 7,17,37 Demultiplexer 8,18 Inverse Quantizer (inverse quantization means) 9 Inverse scaling circuit 10, 20, 40 Band synthesis filter 12, 32 Frame processing circuit (frame processing means) 19, 39 Envelope extraction circuit (envelope extraction means) 21 Zero crossing counting circuit (Zero crossing counting means) 22 Sine wave generation circuit (sine wave generation means) 23, 43 Envelope restoration circuit (envelope restoration means) 24 Mixing circuit (narrow band signal generation means) 25 Difference circuit (difference output means) 34 Zero Crossing interval detection circuit (zero crossing interval calculating means) 35a, 35b Vector quantizer 38a, 38b Vector dequantizer 41 Interpolation point calculation circuit (interpolation point calculation means) 42 Zero crossing interval restoration circuit (zero crossing interval restoration means) 44 Interpolation circuit (interpolation means)

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification number Internal reference number FI Technical indication H03M 7/30 9382-5K H03M 7/30 A

Claims (10)

[Claims] 1. An audio signal to be supplied is divided into a plurality of bands, the band-divided narrowband signals are cut out into short-time analysis frames, and an envelope curve is formed for each frame of each of the narrowband signals. A method for encoding an acoustic signal, which comprises extracting and calculating the number of zero crossings of the amplitude, and encoding the envelope and the number of zero crossings obtained for each frame of each of the narrowband signals. 2. An acoustic signal decoding method for decoding an acoustic signal encoded by the acoustic signal encoding method according to claim 1, wherein the supplied signal is inversely quantized to obtain an envelope and the number of zero crossings. And generate a sine wave based on the number of zero crossings,
A method for decoding an acoustic signal, characterized in that the sine wave is shaped by the envelope to restore a short-time frame, and then an acoustic signal is decoded by synthesizing each frame and each band. 3. The supplied acoustic signal is divided into a plurality of bands, the band-divided narrowband signals are cut into short-time analysis frames, and an envelope curve is formed for each frame of the narrowband signals. A method for encoding an acoustic signal, which comprises extracting and calculating a zero crossing interval of amplitude, and coding the envelope and the zero crossing interval obtained for each frame of each of the narrowband signals. 4. An acoustic signal decoding method for decoding an acoustic signal encoded by the acoustic signal encoding method according to claim 3, wherein the supplied signal is dequantized to generate an envelope and a zero crossing interval. Is restored, the interpolation position on the envelope is calculated based on the zero-crossing interval, and a short-time frame is restored by interpolating the zero-crossing point and the interpolation position on the envelope with a smooth curve. An acoustic signal decoding method characterized by decoding an acoustic signal by synthesizing a frame and each band. 5. A band division filter for dividing the supplied acoustic signal into a plurality of bands, frame processing means for dividing each narrow band signal divided by the band division filter into short time frames, and this frame. Analyzing means for extracting and analyzing characteristics by simplifying the waveform of each of the narrow band signals output from the processing means for each short time frame, and encoding for encoding and transmitting the characteristics extracted and analyzed by the analyzing means An audio signal encoding device comprising: 6. The analyzing means comprises zero-crossing counting means for measuring the number of zero-crossings of the amplitude waveform, and envelope extraction means for extracting the envelope of the amplitude waveform. 5. The acoustic signal encoding device according to item 5. 7. The analyzing means comprises a zero-crossing interval calculating means for calculating a zero-crossing interval of the amplitude waveform and storing it as a time series, and an envelope extracting means for extracting an envelope curve of the amplitude waveform. The acoustic signal encoding device according to claim 5, wherein 8. The acoustic signal encoding apparatus according to claim 6 or 7, wherein the analyzing unit outputs the envelope output from the envelope extracting unit and a short-circuit of the adjacent narrow band signal at the same time. An acoustic signal coding apparatus, characterized by comprising difference output means for outputting a difference from an envelope of a time frame. 9. An acoustic signal decoding apparatus for decoding an acoustic signal encoded by the acoustic signal encoding apparatus according to claim 5 or 6, wherein the inverse quantum is for inversely quantizing the encoded signal. Quantization means, sine wave generation means for generating a sine wave from the number of zero crossings supplied by the dequantization means, envelope restoration means for restoring the envelope curve supplied by the dequantization means, A narrow band signal generating means for generating a narrow band signal from a sine wave supplied from a wave generating means and an envelope supplied from the envelope restoring means; and a narrow band signal supplied from the narrow band signal generating means. An audio signal decoding device, comprising: a band synthesizing filter for synthesizing and outputting an acoustic signal. 10. An acoustic signal decoding apparatus for decoding an acoustic signal encoded by the acoustic signal encoding apparatus according to claim 5 or 7, wherein the inverse quantum is for inversely quantizing the encoded signal. Quantization means, zero crossing interval restoring means for restoring the zero crossing interval from the time series of zero crossing intervals supplied by the dequantizing means, and envelope restoration for restoring the envelope supplied by the dequantizing means. Means, interpolation point calculating means for calculating an interpolation position on the envelope supplied by the envelope restoring means according to the zero-crossing interval supplied by the zero-crossing interval restoring means, and envelope supplied by the interpolation point calculating means Interpolation means for generating a smooth curve passing through the interpolation position on the line and the zero-crossing position calculated from the zero-crossing interval, and band synthesis for synthesizing the narrow band signals supplied from this interpolation means and outputting an acoustic signal. Acoustic signal decoding apparatus characterized by comprising a filter.