Introduction
The latest development in the digital versatile disc (DVD) family is DVD-Audio. As shown in FIG. 1, part 1 and part 2 of the 'DVD Specifications for Read-Only Disc' form the foundation for DVD-ROM. Together with the physical specifications (part 1) and the file system specifications (part 2), part 4 was also recently published to define the specifications for audio applications. Of course, according to part 3 (Video specifications) published before, video information can also be recorded on DVD-Audio.
Inspired by the music industry to create a high-quality recording and archiving system, the DVD Forum Audio Working Group 4, chaired by JVC and composed largely, but not exclusively, of consumer electronics hardware manufacturers, developed the DVD-Audio format.
FIG. 1 DVD specifications for DVD-Audio.
Basic concept
Based on the recommendations of the International Steering Committee (ISC; also see Section 18) and making use of the storage capacity of 4.7 GB on the high-density layer of the DVD-ROM, DVD-Audio is a high-quality digital audio medium.
Building on the knowledge of linear PCM encoding for the conventional CD (limited to a maximum bandwidth of 20 kHz), DVD-Audio uses a sampling frequency of up to 192 kHz.
According to the Nyquist theorem explained in Section 2 of this book, this sampling frequency allows a bandwidth from DC to 96 kHz. In addition to the high sampling frequency, a 24-bit quan tization is used giving a high signal-to-noise ratio.
The calculation of the theoretical signal-to-noise level (S/N) will thus be (refer to Section 3):
S/N (dB) = 6.02 × n + 1.76
where n is the number of quantization bits.
In the case of a 24-bit system, the theoretical S/N ratio will be about 146 dB.
Besides these audio specifications it was also required to include multi-channel possibilities. DVD-Audio is capable of reproducing six-channel digital surround sound.
DVD versus CD
In the comparison given in Table1, the main differences between both formats can be observed.
Compression
Taking into account that at a maximum sampling frequency and 24-bit resolution, the transfer rate becomes 9.6 Mbps (two channels, 192 000 samples per second at 24 bits per sample + error correction, identification and synchronization). At this rate, the maximum recording time on a single-layer HD disc (which has a storage capacity of 4.7 GB) will become less than 74 minutes.
To fit 74 minutes of music onto one disc, a lossless compression method, Meridian Lossless Packing (MLP), has been developed.
Within MLP, three methods are used to reduce the data rate.
Lossless waveform prediction is used to reduce inter-sample correlations, with a very large variety of special filters. In simple terms, this means that comparisons are made between a number of subsequent samples, and where possible, only the differences are encoded. Lossless processing and lossless matrixing are also used to reduce correlations between channels; similar to the inter sample correlation method, here also a comparison is made, but now between channels. Hufman coding is used to reduce the data rate by efficiently encoding the most likely occurring successive values in the serial stream.
Table 1. Comparison between DVD-Audio and CD
FIG. 2 MLP encoder.
MLP reduces the total data amount to about 50% of the original without losing 1 bit after decoding. In practice, MLP is capable of handling up to 63 channels with sampling frequencies between 32 and 192 kHz, so it will perfectly match the requirements of DVD-Audio with maximum six channels and a maximum sampling frequency of 192 kHz (in two-channel mode).
It becomes clear that in the case of six-channel surround sound, the same high-quality standards cannot be achieved for all channels. If six-channel sound were encoded at 192 kHz/24 bits, an audio stream of 27.648 Mbit s^-1 would be necessary to transfer the data, surpassing the limit of 9.6 Mbit s-1 defined for DVD even with MLP. In order to keep the playing time and bit rate to an acceptable level, a mixture of different sampling frequencies and resolutions can be used together with MLP compression.
For example, for a 5.1 surround sound, the front L, front R and centre channel can be encoded in 96 kHz/24 bits, where the left surround, right surround and LFE can be encoded in 48 kHz/16 bits.
Table 2 Different sampling frequencies and corresponding times on a single-layer
disc at 24-bit resolution
FIG. 3 MLP decoder.
Table 2 gives an overview of different sampling frequencies and the corresponding time on a single-layer disc at 24-bit resolution.
FIG. 4 Visual structure of DVD-Audio.
Disc structure
The structure of a DVD-Audio is comparable to conventional CD except that, besides the tracks and indexes (both maximum 99), the album is also divided into a maximum of nine groups. Each group is a collection of a number of tracks. Each group or track is directly accessible by the user, making it very easy to navigate.
Optical read-out system
The optical block used for DVD-Audio is the same as for SACD.
Tracking and focus servo are the same as explained in Section 18.
Options
Other requirements from the ISC were the possibility of including video or text and to create a reliable anti-piracy system.
Based on the DVD specifications for read-only disc parts 1-3, video and text can easily be included.
To meet the anti-piracy requirement, reserved fields to set control flags for copy-protection management are provided. It should be noted that some controversy on this subject has indeed delayed the final opening and use of DVD-Audio. For this reason, no details are given at this point.