Digital Audio Communications Nikil Jayant Bell Laboratories, Lucent Technologies 39 Auditory Psychophysics for Coding Applications Joseph L.. Tsutsui Introduction •Oversampling AD and DA
Trang 1Digital Audio
Communications
Nikil Jayant
Bell Laboratories, Lucent Technologies
39 Auditory Psychophysics for Coding Applications Joseph L Hall
Introduction •Definitions•Summary of Relevant Psychophysical Data•Conclusions
40 MPEG Digital Audio Coding Standards Peter Noll
Introduction •Key Technologies in Audio Coding•MPEG-1/Audio Coding•MPEG-2/Audio
Multichannel Coding •MPEG-4/Audio Coding•Applications•Conclusions
41 Digital Audio Coding: Dolby AC-3 Grant A Davidson
Overview •Bit Stream Syntax•Analysis/Synthesis Filterbank•Spectral Envelope•Multichannel
Coding •Parametric Bit Allocation•Quantization and Coding•Error Detection
42 The Perceptual Audio Coder (PAC) Deepen Sinha, James D Johnston, Sean Dorward,
and Schuyler R Quackenbush
Introduction •Applications and Test Results•Perceptual Coding•Multichannel PAC•Bitstream
Formatter •Decoder Complexity•Conclusions
43 Sony Systems Kenzo Akagiri, M.Katakura, H Yamauchi, E Saito, M Kohut, Masayuki
Nishiguchi, and K Tsutsui
Introduction •Oversampling AD and DA Conversion Principle•The SDDS System for Digitizing
Film Sound •Switched Predictive Coding of Audio Signals for the CD-I and CD-ROM XA Format
•ATRAC (Adaptive Transform Acoustic Coding) and ATRAC 2
A S WE ENTER THE 21ST CENTURY, digital audio communications will have become nearly
as prevalent as digital speech communications In particular, new technologies for audio storage and transmission will make available music and wideband signals in a flexible variety
of standard formats
The fundamental underpinning for these technologies is audio compression based on perceptually-tuned shaping of the quantization noise The next chapter in this section describes psychoacoustics
knowledge that suggests the general principles of perceptual audio coding Succeeding chapters in this section are devoted to descriptions of established examples of perceptual audio coders These include
MPEG standards, and coders developed by Dolby, Sony, and Bell Laboratories
Trang 2The dimensions of coder performance are quality, bit rate, delay, and complexity The quality vs bit rate tradeoffs are particularly important
Audio Quality
The three parameters of digital audio quality are signal bandwidth, fidelity and spatial realism.
Compact-disk (CD) signals have a bandwidth of 20–20,000 Hz, while traditional telephone speech has a bandwidth of 200–3400 Hz Intermediate bandwidths characterize various grades of wideband speech and audio, including roughly defined ranges of quality referred to as AM radio and FM radio quality (bandwidths on the order of 7–10 and 12–15 kHz, respectively)
In the context of digital coding, fidelity refers to the level of perceptibility of quantization or reconstruction noise The highest level of fidelity is one where the noise is imperceptible in formal listening tests Lower levels of fidelity are acceptable in some applications if they are not annoying, although in general it is good practice to sacrifice some bandwidth in the interest of greater fidelity, for a given bit rate in coding Five-point scales of signal fidelity are common in both speech and audio coding
Spatial realism is generally provided by increasing the number of coded (and reproduced) spatial
channels Common formats are 1-channel (mono), 2-channel (stereo), 5-channel (3 front, 2 rear),
5.1-channel (5-channel plus subwoofer) and 8-channel (6 front, 2 rear) For given constraints on bandwidth and fidelity, the required bit rate in coding increases as a function of the number of channels; but the increase is slower than linear, because of the presence of interchannel redundancy The notion of perceptual coding originally developed for exploiting the perceptual irrelevancies of a single-channel audio signal extends also to the methods used in exploiting interchannel redundancy
Bit Rate
The CD-stereo signal has a digital representation rate of 1406 kilobits per second (kb/s) Current technology for perceptual audio coding reproduces CD-stereo with perfect fidelity at bit rates as low
as 128 kb/s, depending on the input signal CD-like reproduction is possible at bit rates as low as
64 kb/s for stereo channel reproduction of FM-radio-like music is possible at 32 kb/s Single-channel reproduction of AM-radio-like music and wideband speech is possible at rates approaching
16 kb/s for all but the most demanding signals Techniques for so-called “pseudo-stereo” can provide additional enhancement of digital single-channel audio
Applications of Digital Audio
The capabilities of audio compression have combined with increasingly affordable implementa-tions on platforms for digital signal processing (DSP), native signal processing (NSP) in a computer’s (native) processor, and application-specific integrated circuits (ASICs) to create revolutionary ap-plications of digital audio International and national standards have contributed immensely to this revolution Some of these standards only specify the bit-stream syntax and decoder, leaving room for future, sometimes proprietary, enhancements of the encoding algorithm
The domains of applications include transmission (for example, digital audio broadcasting), storage (for example, the minidisk and the digital versatile disk, DVD), and networking (music preview,
distribution, and publishing) The networking applications will make digital audio communications
as commonplace as digital telephony
The Future of Digital Audio
Remarkable as the capabilities and applications mentioned above are, there are even greater chal-lenges and opportunities for the practitioners of digital audio technology It is unlikely that we have reached or even approached the fundamental limits of performance in terms of audio quality at
a given bit rate Newer capabilities in this technology (in terms of audio fidelity, bandwidth, and
Trang 3spatial realism) will continue to lead to newer classes of applications in audio communications New technologies for embedded coding and universal coding will create interesting new options for digital networking and seamless communication of speech and music signals Finally, co-designs of audio processing with image and video processing will lead to currently unavailable capabilities for multi-media networking games, computer agents, and personal communication services These scenarios will call upon our best capabilities in signal compression as well as advances in the sister disciplines
of signal synthesis and recognition by machine