Subband encoding and decoding system

Details Subband encoding and decoding system Subband encoding and decoding system

2000-07-26

2002-09-24

Dorvil, Richemond (Department: 2654)

Data processing: speech signal processing, linguistics, language

Speech signal processing

For storage or transmission

C704S500000, C704S501000, C704S230000

Reexamination Certificate

active

06456968

ABSTRACT:

BACKGROUND OF THE INVENTION
The present invention relates to a subband encoding and decoding system, and more particularly to a subband decoding and decoding system preferably used for compressing a digital signal.
A representative example of a conventional subband encoding system is a MPEG1 audio system.
FIG. 40
is a block diagram showing a conventional MPEG1 audio layer encoding system. An encoding input digital signal s
101
of sampling frequency fs, which is an input signal entered into this encoding system, is supplied to a band splitting section a
101
. The band splitting section a
101
splits the input signal s
101
for encoder into a total of k band components successive in an entire frequency zone ranging from 0 to a Nyquist frequency (fs/2) of the encoder input signal s
101
, where “k” is an arbitrary integer. The band splitting section a
101
outputs each subband signal s
102
of k split bands. For example, MPEG1 audio is based on a uniform band width slpitting of k=32. However, instead of using the uniform splitting, it is possible to adopt a non-uniform splitting depending on an individual filter arrangement, provided that each of k split band widths is a predetermined fixed value.
Furthermore, in the MPEG1 audio layer, each subband signal is down-sampled into a baseband signal by using a sort of frequency modulation. Meanwhile, while maintaining time synchronization with the band splitting section a
101
, a time-frequency converting section a
102
performs a time-frequency conversion on the encoding input digital signal s
101
, wherein time window curtain is applied to w samples each having a unit sample length equivalent to a reciprocal (1/fs) of the sampling frequency. The time-frequency converting section a
102
outputs frequency information s
103
as a result of the time-frequency conversion. A time window length “w” used for the time-frequency conversion is obtained according to a frequency resolution “fr” required for the frequency information s
103
.
W
=(1
/fr
)/(1
/fs
)
According to the MPEG1 audio layer, a fast Fourier transform technique is used for time-frequency conversion. The value “w” is defined as a minimum value 2
w
satisfying the required frequency resolution “fr.” Furthermore, considering time continuity, an appropriate overlap zone is provided between two consecutive time windows.
A frequency analyzing section a
103
calculates a bit allocation number for each of k split bands to produce bit allocation information s
104
, by using a conventionally known auditory masking based on a psychoacoustic model, during a time length of a time window excluding the overlap zone used in the time-frequency converting section a
102
. The time length of a time window excluding the overlap zone is a unit time length of frame. An encoding section a
104
produces a scale factor of each split band with reference to a maximum amplitude value per unit frame length of each subband signal s
102
. Based on the obtained scale factor of each split band, the amplitude of each subband signal s
102
is normalized. Subsequently, requantization for each split is performed band based on the bit allocation information s
104
. The encoding section a
104
forms a bit stream including the requantized sample, the bit allocation information, the scale factor and a frame sync information. Thus, the encoding section a
104
produces a coded output signal s
105
.
FIG. 41
is a block diagram showing a conventional MPEG1 audio layer decoding system. A decoder input signal s
106
, which is a coded signal produced from the encoding system, is entered into the decoding system. A frame analyzing section a
105
detects a frame, bit allocation information, and scale factor contained in the decoder input signal s
106
, thereby producing frame analysis information s
107
. A decoding section a
106
performs the decoding processing for each split band based on the frame analysis information s
107
to output a subband signal s
108
. Subsequently, a band combining section a
107
combines the subband signals s
108
to output a decoded output signal s
109
. To prevent any deterioration of information through the encoding-decoding processing, the condition required for the band combining device is to establish perfect reconstruction conditions matching with the band splitting section a
101
of the encoding system. A conventionally known technique using QMF provides a filter arrangement satisfying such perfect reconstruction conditions.
However, the conventional MPEG subband encoding system performs the scale factor information and bit allocation information producing processing as well as the requantization processing for each of k split bands, and then constructs a frame with reference to the obtained information. This significantly increases a processing amount in the encoding processing and also increases a bit rate.
Furthermore, the conventional MPEG subband encoding system performs the compression of information based on the psychoacoustic model. Thus, the time-frequency conversion and the signal analysis in frequency regions are inevitable. To realize highly efficient compression without causing deterioration of information, it is necessary to maintain sufficient frequency resolution. To realize this, the frequency conversion requires a window curtain applied to a sufficiently long time sample. Regarding the delay time in the subband encoding and decoding processing, a frame length is determined based on a sample number required for the window curtain processing. This frame length serves as a unit length for performing each of the encoding processing, the decoding processing, and the buffering processing. Thus, the delay time depends on a processing time per frame length caused in each processing and a group delay of the band split filter. Thus, the processing delay time necessarily increases when to realize high sound quality and high compression rate.
Furthermore, the conventional MPEG subband encoding system requires a great processing amount for the frequency analysis and the bit allocation processing.
Moreover, when the conventional MPEG subband encoding system is used for radio transmission, it is necessary to add the sync word generating processing and the sync word detecting processing for performing a clock sync acquisition of a receiving system and a synchronization of a radio frame. To reduce errors caused in a transmission path, it is necessary to add the error correction processing separately. Accordingly, the processing delay time of an overall system further increases due to a buffering time in each processing etc. The separately added error correction processing is performed without considering characteristics of each information in the subband encoding processing. Thus, even in a preferable condition where a burst error is acceptable or a bit error rate during a long time period is relatively low, a fatal error may occur at an application level.
SUMMARY OF THE INVENTION
To solve the above-described conventional problems, the present invention has an object to provide a subband encoding and decoding system capable of reducing both the encoding processing amount and the encoding bit rate.
In order to accomplish this and other related objects, the present invention provides a subband encoding system comprising a band splitting means for implementing a band splitting on an encoder input signal to produce subband signals, a scale factor producing means for producing scale factor information in accordance with a signal output level of each subband signal, a bit allocation producing means for calculating bit allocation information based on the scale factor information, a requantizing means for implementing requantization based on the subband signal, the scale factor information and the bit allocation information, thereby outputting a requantized output signal, a frame constructing means for constructing a coded frame based on the requantized output signal and the scale factor information, thereby outputting a coded output signal, and a limiting means for limiting a tot

Affiliated with

Also associated with

Rating

Subband encoding and decoding system does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Subband encoding and decoding system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Subband encoding and decoding system will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2863392