Data processing: speech signal processing – linguistics – language – Speech signal processing – For storage or transmission
Reexamination Certificate
1999-08-18
2004-07-20
{haeck over (S)}mits, T{overscore (a)}livaldis Ivars (Department: 2654)
Data processing: speech signal processing, linguistics, language
Speech signal processing
For storage or transmission
C704S500000, C704S227000
Reexamination Certificate
active
06766293
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to audio coding methods and in particular to audio coding methods according to the Standard ISO/MPEG, such as e.g. MPEG-1, MPEG-2, MPEG-2 AAC, for the data-reduced representation of high quality audio signals.
BACKGROUND OF THE INVENTION AND DESCRIPTION OF PRIOR ART
The standardization body ISO/IEC JTC1/SC29/WG11, which is also known as the Moving Pictures Expert Group (MPEG), was founded in 1988 in order to specify digital video and audio coding schemes for low data rates. In November 1992 the first specification phase was completed with the Standard MPEG-1. The audio coding system according to MPEG-1, which is specified in ISO 11172-3, works in a one-channel or two-channel stereo mode at sampling frequencies of 32 kHz, 44.1 kHz and 48 kHz. The Standard MPEG-1 Layer II delivers radio quality, as it is specified by the International Telecommunication Union, at a data rate of 128 kb/s per channel.
In its second development phase the aims of MPEG were to define a multichannel extension for MPEG-1 audio which should be backwards compatible with the existing MPEG-1 systems, and also to define an audio coding standard at lower sampling frequencies (16 kHz, 22.5 kHz, 24 kHz) than in MPEG-1. The backwards compatible standard (MPEG-2 BC) and the standard with lower sampling frequencies (MPEG-2 LSF) were completed in November 1994. MPEG-2 BC delivers a good audio quality at data rates of 640-896 kb/s for 5 channels with full bandwidth. Since 1994 the MPEG-2 Audio Standardization Committee has been striving to define a multichannel standard with higher quality than is attainable if backwards compatibility with MPEG-1 is required. This non-backwards-compatible audio standard according to MPEG-2 is denoted by MPEG-2 NBC. The aim of this development is to achieve radio quality according to the ITU-R requirements at data rates of 384 kb/s or less for 5-channel audio signals for which each channel has the full bandwidth. The Audio Coding Standard MPEG-2 NBC was completed in April 1997. The scheme MPEG-2 NBC will become the nucleus of the already planned Audio Standard MPEG-4, which will have higher data rates (over 40 kb/s per channel). The NBC or non-backwards compatible standard combines the coding efficiency of a high-resolution filter bank, of prediction techniques and of the redundancy reducing Huffman coding to achieve an audio coding with radio quality at very low data rates. The Standard MPEG-2 NBC is also denoted by MPEG-2 NBC AAC (AAC=Advanced Audio Coding). A detailed description of the technical content of MPEG-2 AAC is to be found in M. Bosi, K. Brandenburg, S. Quackenbush, L. Fiedler, K. Akagiri, H. Fuchs, M. Dietz, J. Herre, G. Davidson, Yoshiaki Oikawa: “ISO/IEC MPEG-2 Advanced Audio Coding”, 101st AES Convention, Los Angeles 1996, Preprint 4382.
Efficient audio coding methods remove both redundancies and irrelevancies from audio signals. Correlations between audio sampling values and statistics of the sampling value representation are exploited so as to remove redundancies. Frequency domain and time domain masking properties of the human auditory system are exploited so as to remove imperceptible signal content (irrelevancies). The frequency content of the audio signal is subdivided into subbands by means of a filter bank. The data rate reduction is achieved by quantizing the spectrum of the time-domain signal according to psychoacoustic models and may include a lossless coding method.
Generally speaking, a time-continuous audio signal is sampled so as to obtain a time-discrete audio signal. The time-discrete audio signal is windowed by means of a window function so as to obtain successive blocks or frames with a certain number, e.g. 1024, of windowed time-discrete sampled values. Each block of windowed time-discrete sampled audio signal values is transformed in turn into the frequency domain, which may be achieved using a modified discrete cosine transform (MDCT) for example. Since the spectral values obtained in this way are not yet quantized, it is necessary to quantize them. Here the main aim is to quantize the spectral data in such a way that the quantization noise is masked or concealed by the quantized signals themselves. This is achieved with the aid of a psychoacoustic model described in the MPEG AAC Standard which, taking account of the special properties of the human ear, calculates masking thresholds depending on the audio signal involved. The spectral values are now quantized in such a way that the quantized noise which is introduced is concealed and therefore inaudible. The quantization does not therefore result in any audible noise.
In the NBC Standard a so-called non-uniform quantizer is used. Additionally, a method for shaping the quantization noise is used. The NBC method, like previous standards, employs the individual amplification of groups of spectral coefficients, which are known as scale factor bands. To work as efficiently as possible it is desirable to be able to shape the quantization noise into units which are based as closely as possible on the frequency groups of the human auditory system. In this way it is possible to group together spectral values which very closely reflect the bandwidth of the frequency groups. Individual scale factor bands can be amplified by means of scale factors in stages of 1.5 dB. The noise shaping is achieved since amplified coefficients have larger amplitudes. They will therefore in general have a higher signal
oise ratio after quantization. On the other hand, larger amplitudes require more bits for the coding, i.e. the bit distribution between the scale factor bands is implicitly changed. The amplification through the scale factors must of course be corrected in the decoder. For this reason the amplification information, which is stored in the scale factors in units of 1.5 dB steps, must be transmitted to the decoder as side information.
After quantization of the spectral values, possibly amplified through scale factors, in the scale factor bands, the spectral values themselves should be coded. The input signal into a noiseless coding module is thus the set of e.g. 1024 quantized spectral coefficients. The sets of 1024 quantized spectral coefficients are partitioned by the noiseless coding module into “sections” in such a way that a single Huffman codebook is used to code each section. For reasons of coding efficiency, section boundaries can only exist at scale factor band boundaries such that for each section of the spectrum both the length of the section in scale factor bands and the Huffman codebook number used for the section must be transmitted as side information.
The forming of the sections is dynamic and varies typically from block to block in such a way that the number of bits needed to represent the full set of quantized spectral coefficients is minimized. The Huffman coding is used to represent n-tuples of quantized coefficients, the Huffman code being derived from one of 12 codebooks. The maximum absolute value of the quantized coefficients which can be represented by each Huffman codebook and the number of coefficients in each n-tuple for each codebook are specified a priori.
The point of forming the sections thus consists in grouping together regions with the same signal statistics so as to obtain, with a single Huffman codebook for a section, the highest possible coding gain, the coding gain generally being defined as the quotient of the bits before coding and the bits after coding. By means of a codebook number, which is specified in the bit stream syntax used for the NBC method, one of the 12 Huffman codebooks is referred to, namely the one which makes possible the highest coding gain for a specific section. The expression “codebook number” in this application is thus meant to designate the place in the bit stream syntax which is reserved for the codebook number. To code 11 different codebook numbers in binary, 4 bits are required. For each section, i.e. for each group of spectral values, these 4 bits must be transmitted as side informatio
Brandenburg Karlheinz
Dietz Martin
Ehret Andreas
Gbur Uwe
Gerhäuser Heinz
Armstrong Angela
Fraunhofer-Gesellschaft Zur Foerderung der Angewandten Forschung
Glenn Michael A.
Glenn Patent Group
{haeck over (S)}mits T{overscore (a)}livaldis Ivars
LandOfFree
Method for signalling a noise substitution during audio... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for signalling a noise substitution during audio..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for signalling a noise substitution during audio... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3193309