Back
Next
Encoding data on a CD ISO 9660
see wikipedia : compact disc
Original protocol : 650 MiB data or 74 minutes audio.
Improvements in cd readers allow for more "tracks"
New standard : 700 MiB data or 80 minutes audio.
With 800 MiB/90 minutes or 900 MiB/99 minutes available but custom.
The wobble track has to be correctly positioned, so you can't just
burn more 'tracks' on a standard CD-R
Data recorded with NRZI timing and uses 8 to 14 bit (2,10 RLL) encoding.
Initial sector level data type.
Audio
Sound sampled 2 x highest frequency. 44,100 Hz (A to D)
(60 Hz - 20,000 Hz human hearing range)
16 bit signed value for each sample.
(Analog recording vs. digital recording - wikipedia)
6 samples from left and right (12 total) = 24 byte frame.
The series of frames is subjected to Cross-interleaved Reed-Solomon
encoding which adds a parity byte for every three bytes of data.
And a long series of these frames are 'scrambled' or interleaved
to guard against burst errors.
An additional 8 bits, call sub-channel data is added to each frame.
Song information stored in the sub-channel bytes of multiple frames.
Total frame size is 33 bytes called channel-data.
The channel-data is then encoded in 8,14 RLL and 3 merging bits are
added before each 14-bit byte.
A 27-bit word is added to start of each frame.
# Note that this gives 588 bits to encode 192 bits of 'sound'.
The frames are then collected into sectors called Time Code Frames (TCF).
98 channel-frames make a timecode frame.
TCFs are stored on CD sequentially.
No TCF level error protection.
* 2352 bytes of data.
* No sector ID.
* No dedicated error correcting or checking block.
Audio is addressed as a continuous recording with very crude
rapid search for tracks (songs).
Read sequentially.
Played at 75 frames a second for audio.
Tracks or songs are found by reading the sub-channel data.
Note that audio files are not the same as MP3 or wav files which are
sampled and encoded as bytes of data in as data file format.
MP3 and wav files are made with additional loss of sound quality,
but there is a debate as to whether that is actually perceptible.
The understanding of audio data sampling has improved greatly over
the years.
Digital modes - still use the framing structure.
Digital (mode 1) - data. All bits saved.
Includes sector id and error correcting data.
Sector level ECC similar to hard drives.
* 12 byte sync
* 3 byte sector address - different from LBA because it is actual block @
* 1 byte mode flag
* 2048 bytes data
* 4 byte error detection
* 8 byte reserved, zero
* 276 byte ECC block.
Digital video. (mode 2) Sampling and encoding.
Mixed audio/video and data, some error tolerated.
14% more storage than data mode 1
Designed to be addressed by sectors, contains header info but
no error correcting.
* 12 byte sync
* 3 byte address(sector ID)
* 1 byte mode flag
* 2336 bytes data
Digital XA (CD-ROM XA) - extended architecture (for mode 2).
Allows for mixing of data and video sectors with their different
formats. Including compressed data.
* may be obsolete.
(Form 1)
* 12 byte sync
* 3 byte address(sector ID)
* 1 byte mode flag
* 8 byte (sub-header)
* 2048 bytes data
* 4 byte error detection
* 276 byte ECC block.
# almost identical to Mode 1 (data) but disc flagged CD-ROM XA for
devices that expect to see such discs.
(Form 2 - for video, synced time more critical than accuracy)
* 12 byte sync
* 3 byte address(sector)
* 1 byte mode flag
* 8 byte (sub-header)
* 2324 bytes data
* 4 byte error detection
Note : because of improved density and speed of access, DVDs and Blu-Ray
did away with multiple alternative modes, data is stored as files using
a variation of CD Mode 1.
Header block
Data block
Error correction block.