Coded data generation or conversion – Digital code to digital code converters
Reexamination Certificate
2000-12-04
2003-02-11
Jeanpierre, Peguy (Department: 2819)
Coded data generation or conversion
Digital code to digital code converters
C341S051000, C704S230000
Reexamination Certificate
active
06518891
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an encoding apparatus and method, adapted to encode a second code string conforming to a second format based on a second coding method with a higher efficiency than that of a first code string conforming to a first format based on a first coding method.
2. Description of the Related Art
The technique to record information to a recording medium capable of recording an encoded audio or speech signal, such as a magneto-optical disc or the like, is widely used. For a highly efficient coding of an audio or speech signal, there have been proposed various methods such as the subband coding method (SBC) in which an audio signal or the like on a time base is divided into a plurality of frequency bands without blocking, and the so-called transform coding method in which a signal on the time base is transformed to a signal on the frequency base (spectrum transform), divided into a plurality of frequency bands, and then the signal in each of the frequency bands is encoded. Also, a high efficiency coding method has also been proposed which is a combination of the SBC method and transform coding method. In this third method, for example, after an audio or speech signal is divided into a plurality of frequency bands by the SBC method, the signal in each frequency band is spectrum-transformed to a signal on the frequency base, and the signal is encoded in each spectrum-transformed frequency band. The QMF filter is defined in R.E. Crochiere: “Digital Coding of Speech Subbands”, Bell Syst. Tech. Journal, Vol. 55, No. 8, 1976″. Also, the method for equal-bandwidth division by filter is defined in Joseph H. Rothweiler: “Polyphase Quadrature Filters—A New Subband Cording Technique”, ICASSP 83, BOSTON.
In an example of the above-mentioned spectrum, an input audio signal is blocked at predetermined unit times (frames), and each of the blocks is subjected to the discrete Fourier transform (DFI), discrete cosine transform (DCT) or modified discrete cosine transform (MDCT) to transform a time base to a frequency base. The MDCT is described in “J. P. Princen and A. B. Bradley, Univ. of Surrey Royal Melbourne Inst. of Tech.: Subband/Transform Coding Using Filter Bank Designs Based on Time Domain Aliasing Cancellation, ICASSP, 1987”.
When the above-mentioned DFT or DCT is used for transform of a waveform signal to a spectrum, with a time block consisting of M samples will yield a number M of independent real data. Normally, a time block is arranged to overlap M
1
samples of its neighboring blocks each to suppress the distortion of the connection between time blocks. Therefore, in the DFT and DCT, a signal will be encoded by quantizing on average M real data for a number (M-M
1
) of samples.
When the MDCT is used as the method for transform of a waveform signal to a spectrum, M independent real data can be obtained from
2
M samples arranged to overlap M ones of its neighboring blocks each. Therefore, in the MDCT, the signal is encoded by quantizing on average M real data for the M samples. In a decoder, waveform elements obtained from a code resulted from the MDCT by inverse transform in each block are added together while being made to interfere with each other, thereby permitting reconstruction of the waveform signal.
Generally, by increasing the length of the time block, the frequency separation of the spectrum is increased and energy is concentrated on a specific spectrum component. Therefore, by transforming a waveform signal to a spectrum with an increased block length obtained by overlapping a time block a half of its neighboring time blocks each and using the MDCT in which the number of spectrum signals obtained will not increase relative to the number of original time samples, it will be possible to enable a coding whose efficiency is higher than that attainable with the DFT or DCT.
By quantizing a signal divided into a plurality of frequency bands by the filtering or spectrum transform as in the above, it is possible to control any frequency band where quantization noise occurs and encode an audio signal with a higher efficiency in the auditory sense, using a property such as the masking effect. Also, by normalizing, for each of the frequency bands, the audio signal with a maximum absolute value of a signal component in the frequency band before effecting the quantization, a further higher efficiency of the coding can be attained.
The width of frequency division for quantization of each frequency component resulted from a frequency band division is selected with the auditory characteristic of the human being for example, taken into consideration. That is, an audio signal is divided into a plurality of frequency bands (25 bands for example) in such a bandwidth as will be larger as its frequency band is higher, which is generally called a “critical band”, as the case may be. Also, at this time data in each band is encoded by a bit distribution to each band or with an adaptive bit allocation to each band. For example, when a coefficient data obtained using MDCT is encoded with the above bit allocation, an MDCT coefficient data in each band, obtained using the MDCT at each block, will be encoded with an adaptively allocated number of bits. The adaptive bit allocation information can be determined so as to be previously included in a code string, whereby the sound quality can be improved by improving the coding method even after determining a format for decoding. The known bit allocation techniques include the following two:
One of them is disclosed in “R. Zelinski and P. Noll: Adaptive Transform Coding of Speech Signals”, IEEE Transaction of Acoustics, Speech, and Signal Processing, Vol. ASSP-25, No. 4, August 1977. This technique is such that the bit allocation is made based on the size of a signal in each frequency band. With this technique, the quantization noise spectrum can be flat and the noise energy be at a minimum, but since no masking effect is used, the actual noise will not feel auditorily optimum.
The other one is disclosed in “M. A. Kransner, MIT: The Critical Band Coder—Digital encoding of the perceptual requirements of the auditory system, ICASSP, 1980”. This technique is such that the auditory masking is used to acquire a necessary signal-to-noise ratio for each frequency band, thus making a fixed bit allocation. With this technique, however, since the bit allocation is a fixed one , the signal characteristic will not be so good even when it is measured on a sine wave input.
To solve the above problem, there has been proposed a high efficiency encoder in which all bits usable for the bit allocation are divided for a fixed bit allocation pattern predetermined for each small block and for a bit distribution dependent upon a signal size of each block at a ratio dependent upon a signal related with an input signal and whose number of bits for the fixed bit allocation pattern is larger as the spectrum of the signal is smoother.
With the above method adopted in the encoder, the entire signal-to-noise ratio can considerably be improved by allocating more bits to a block including a specific spectrum to which energy is concentrated, such as a sine wave input. Generally, since the human ears are extremely sensitive to a signal having a steep spectrum component, the above method can be used to improve the signal-to-noise ratio, which does not only improve a measured value but also can effectively improve the sound quality.
The bit allocation methods include many other ones as well. The auditory model is further elaborated to enable a higher-efficiency coding if the encoder could. Generally, in these methods, a reference for the real bit allocation to realize a computed signal-to-noise ratio with a highest possible fidelity is determined and an integral value approximate to the computed value is taken as a number of allocated bits.
For example, the present invention has proposed an encoding method in which a signal component having an auditorily important tone component, namely, a signal component ha
Honma Hiroyuki
Miyazaki Satoshi
Shimoyoshi Osamu
Tsutsui Kyota
Jeanglaude Jean Bruner
Jeanpierre Peguy
Sonnenschein Nath & Rosenthal
Sony Corporation
LandOfFree
Encoding apparatus and method, recording medium, and... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Encoding apparatus and method, recording medium, and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Encoding apparatus and method, recording medium, and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3119111