AUdIoCoUrSeS
Joined: 31 Oct 2002
Posts: 2014
|
| Week 5 |
 |
|
Questions for this week
Error handling systems
With this weeks questions i'm looking for much more elongated answers and details. Look to answer these very fully, particularly the last question.
1. What is the sub-code area of the DAT tape used for?
2. Explain the principles of predictive coding.
3. How does the data buffer of a Minidisc player facilitate editing?
4. What are the basic principles of masking?
5. Explain the the following, with reference to a HD editing system
(a) disc access time
(b) EDL
(c) Disc data bandwidth
6. Explain briefly the problems of transferring data files between different proprietors HD systems.
7. Explain the principles used in editing on a single Mini Disc system.
8. Describe and explain the following error handling systems
• Causes of errors
• Minimisation of the consequences of errors
• Error detection
• Error correction
• Error concealment
_________________
It's all in the ears. - Learn the concepts not the software.
Audio Courses is a way into the music business for you
|
Mon Sep 26, 2005 4:14 am |
|
|
rachelh
Joined: 16 Jan 2005
Posts: 35
Location: Trinidad WI |
|
|
|
WEEK 5: ERROR HANDLING SYSTEMS
1. What is the sub-code area of the DAT tape used for?
In DAT, each recorded scan track consists of 196 blocks contained within 16 tracks. Sub-code is contained in two tracks: tracks 3 and 14; each of which is bordered with a PLL [Phase-Locked Loop] and Post amble barrier track. The sub-code data is recorded redundantly to assure accuracy and is written in 8 blocks for each sub code area. The PLL preambles, which is written in 2 blocks are contained in tracks 2 and 13 respectively and contain data, which label fields and lock the data within the fields, which in turn allows for independence in sub code writing.
Each sub code block contains a synchronisation byte, ID code [W1] byte, block address code [W2] byte, parity byte, and 32 data bytes.
The W1 byte contains PCM ID sub code data, which cannot be edited, sampling frequency, quantization level, tape speed, copy-inhibit flag, channel number as well as other parameters are included in this ID.
The W2 block identifies if the data block contains audio or data sub-code data the MSB is 0 for audio and 1 for sub code. In sub code blocks the least significant four bits contain block addresses [0 to 15] the other 3 bits contain sub code ID data.
The parity byte is used to detect errors in the ID and block address bytes, the parity byte itself is the EXCLUSIVE OR sum of the ID and block address bytes.
The remainder of the sub code block contains 256 bits, which equate to 32 data bytes of either audio or sub code data.
The majority of the DAT sub code itself contains data, which is used for program timing, and numbering - called sub code ID and is written along the scan track and in turn providing greater resistance to dropouts. The ID bits contain flags such as Start ID, Skip ID, TOCID [Table of contents ID], Data ID and Format ID information that state how many packs of data have been recorded. Sub code ID data is very complex and flexible in nature and within its 64 bit parameter contains information pertaining to program time, absolute time, running time, table of contents, catalogue number, ISRC [International Standard Recording Code], professional time code, as well as parity data.
In general, the sub code area of DAT allows for data to be recorded alongside the Audio in a DAT track, the data of which pertains to parameters of the audio itself and help to ensure error free and accurate recording and playback.
[1]
2. Explain the principles of predictive coding.
Predictive or perceptive coding refers to codes that are based on the psychoacoustics of sound and hearing. This type of coding relies on the principles of masking – where our auditory perception is less sensitive to sound at one frequency whilst another frequency of close value is being heard. Hence the higher frequency masks the lower frequency. In perceptive coding, the greater the compression factor, the more accurately the human senses must be modelled. And more quantization can take place and will be masked by the greater frequency. So, previously decoded data can be used to predict current data. So, data can be transmitted with omissions and a predictive codec can be used to accurately predict the missing data by examining the previous data values and estimating what the omitted value will be, this value is then subtracted from the post omitted data value and produces a prediction residual error that is transmitted from the encoder to the decoder which in turn interprets this data and produces an output value which is used to replace the omitted data. [2]
3. How does the data buffer of a Minidisc player facilitate editing?
Minidisc [MD] has RAM [Random Access Memory] whose capability allows for about 3 seconds of data/ audio to be buffered. The buffer itself is an area in memory that stores temporary data that is being used for data transfer, the larger the buffer the faster data can be processed. The data rate from MD exceeds what audio decoders need. Procedure wise, when the RAM has reached capacity the disk drive stops sending out data but keeps on oscillating, the data itself it transmitted to the decoders via bursts of data that leaves a ¾ time period of inactivity from the disk. Keeping this in mind, the data buffer of a Minidisc player facilitates editing as this time lapse allows the disk to reposition itself between data transfers so data can be edited. [2]
4. What are the basic principles of masking?
Masking is the process by which our auditory perception is less sensitive to sound at one frequency whilst another frequency of close value is being heard. Hence the higher frequency masks the lower frequency. In other words, if the ear is processing two frequencies and they are close in value, the louder or greater frequency will prevent the ear and brain from interpreting and processing the frequency of lesser value. Masking can also be caused by harmonics of the masking tome. Equalisation can be used to help combat the masking effects. [2]
5. Explain the following, with reference to a HD editing system
(a) Disc access time
In a hard disk editing system the disk access time refers to the time that it takes for the system to access information stored on the disk. Fragmentation, high disk latency, disk speed [in rpm], processor speed and buffer size and memory all affect the disk access time.
Fragmentation is when data stored on the hard drive is broken up into pieces, due to data being moved, re-written or deleted. Fragmentation slows down the disk as a file may be broken up into many pieces and instead of being stored continuously onto adjacent clusters and sectors, thus, the disk head may have to skip over a few sectors to allocate a file in its totality slowing down file retrieval and processing.
Latency describes the actual delay in terms of processing a signal or processing software. Latency therefore is the time taken for a signal to pass through to an electronic device [the time taken for a device to respond to the request / command at hand] or the time taken for a request to be processed and applied. Propagation delay, describes the initial delay that will occur that is, the time it will take the signal to pass through a signal processing box, thus propagation delay contributes to latency, in general, the higher the latency of the device, the longer it would take for processing.
Processor speed, the amount of RAM and Hard Drive Space, all of which contribute to the speed of an operating system are essential to a HD editing system. The amount of RAM namely cache memory is vital as it provides the system with the much-needed capacity to access recently used data, such as clips copied onto the clipboard. Hard Drive Space is essential as the size of a hard drive is directly proportional to the capacity of workload that the system can achieve, when drive space is limited, applications begin to run slower and in severe cases files will not be able to be saved as due to the lack of space. Processor speed also contributes to system effectiveness, the higher the clock speed of the processor; the faster the disk can execute instructions.
[3][4][5]
(b) EDL
EDL or Edit Decision List refers to a list of desired takes and edits that will be used from the master recording, along with notes on where the cuts and edits will be performed. HD systems do not need to edit the actual data files on the disk. Editing is done in the memory of the control system and is dynamically repeated under EDL control each time the edited work is required. Hence, a lengthy editing session on a HD system will not result in the disk becoming fuller as only a few bytes of EDL are generated. When the file is closed off the EDL is stored on disk so that when the file is opened up the EDL will be used to determine what files will be needed and when it will be needed. [6][2]
(c) Disc data bandwidth
The Disc data bandwidth of a HD system is the rate at which data can be processed by the hard disk. Bandwidth is usually expressed in bits or bytes per second. Thus, the higher the bandwidth the faster processes can take place and the lower the latency of the device, so the disc data bandwidth of a HD system determines the writing speed as well as reading speed of the disk.
6. Explain briefly the problems of transferring data files between different proprietors HD systems.
Most HD systems and their applications use session files. Session files are usually not transportable between different systems as there is currently no universal compatibility between systems for this format. Thus, if you take one disk cartridge and load it into a disk recording system that is not the same as the one in which the data was originally stored/recorded, there will be problems with respect to reading the data if it is read at all. Information pertaining to the name of the project, the tracks [how they are assigned], and loops/ audio files used and editing, which occurred, is stored within a session file.
7. Explain the principles used in editing on a single Mini Disc system.
The recordable MiniDisc resembles a hard disk than a real time audio recorder. Data is fragmented across the disc surface and the buffer memory allows for continuous audio listening. The UTOC – User Table Of Contents, makes and lists an entry for each recorded item by noting the physical cluster addresses at which the data for all items are recorded. The UTOC is read in order to locate the address of a track when it is called upon to do so, the item numbers are stored contiguously so if an item is deleted or if two items are merged, the numbering scheme beyond will move up by one. It then becomes unnecessary to actually delete recordings as the UTOC address can be deleted and this in itself will signify that the recording is deleted and the clusters that it uses can be overwritten with new data. Keeping this in mind, editing on a single MiniDisc system is the same as editing on a HD system, tracks can be: deleted, moved, and editing can take place effortlessly and the UTOC is updated in relation to what changes are made. [2]
8. Describe and explain the following error handling systems
• Causes of errors
Errors are caused by either design defects, a malfunctioning device or problems in transmission. Magnetic media can be affected by dust, scratches, fingerprints, tape stretching/ abuse, impure oxide binder, irregular tape setting whilst optical media can be affected by asymmetry, bubbles or defects in the substrate, or coating defects. The severity of the error will depend on which bit is corrupted for instance an error in the LSB would most likely go unnoticed. Errors with no relation to one another are called random bit errors and are easily corrected. Burst errors are large and often affect thousand of bits and are caused by either a malfunctioning defect, a foreign particle on the disc or tape, noise spikes, crosstalk or connector problems in a transmission cable. Dropouts are the most significant cause of errors and are caused by drops in the signal strength and result in corrupted data and may manifest itself as clicks or pops or general noise in the playback. [1]
• Minimisation of the consequences of errors
The minimisation of error consequences would result in the salvaging of work that has been damaged. Errors are considered in the writing process in tape as well as optical media as there is a level of redundancy in data writing and systems such as interleaving have been adopted. Procedures such as interleaving where data is dispersed so that if any section of data is lost, the data, which is dispersed, will contain enough information in which to reconstruct the diminished damaged signal are employed to help minimise the consequences of errors.
• Error detection
Error detection techniques are based on redundancy of data. Redundant data is that which is derived from principal data and conveys no additional information. ‘The task of error detection is to properly code transmitted or stored information so that when the data is lost or made invalid, the presence of the error can be detected’ [1].
Some error detection techniques are as follows:
Single bit parity refers to a method of error detection [not correction] whereby if the number of 1’s in a data word if even or zero the parity bit will be equal to 0; if the number of 1’s in a data word is 1 then the parity bit will be equal to 1. During playback the validity of the received data is checked back using the parity data, which in turn means that the received data bits are added together to calculate the parity of the received data. Errors occur when the received data and the parity bit are not equal. [1]
CCRC or Cyclic redundancy check code is another means of error detection. CCRC is able to detect burst errors in the transmission channel, where burst errors are the most common error type that occurs. CCRC uses a checksum in which data are added together in the form of words, if this checksum and the received data do not add up it is evidence that an error has occurred. In CCRC errors can be corrected by forming an error pattern that is the difference between the received data and the original data to be recovered. [1]
• Error correction
Simple methods of error detection such as redundancy usually are effective but not economical as they use a lot of headspace and in turn may be the source of errors. Error correction codes are formed using redundant data, generally, two types of error codes are used; block codes using algebraic methods and convolutional codes using abilistic methods.
Block coders work by assembling data words to form a block then operating over it parity words are generated and appended to the block. During the decoding process, an algorithm forma a syndrome word that detects the errors and given the redundancy present the errors are corrected.
Hamming Codes are block codes that point the direction of the errors and use multiple parity bits, which are formed for each data word with a unique coding.
Convolutional codes, which are also called recurrent codes, are different from block codes in the way that the data is organised for coding. Convolutional codes do not partition data into blocks. Upon retrieval, the correction decoder uses syndromes to check codewords for errors and due to the redundancy errors can be corrected.
Interleaving, Cross Interleaving, Reed Solomon Codes and Cross Interleave Reed Solomon Codes are other methods of error correction. [1]
• Error concealment
Error concealment is very important if left untreated, errors will audibly noticeable as clicks and pops in the audio. Errors can be concealed by the following methods:
Interpolation, works by identifying the error, and holds the previous correct sample and repeats it to conceal the error [the missing or incorrect sample] – this is called zero-order or pervious sample interpolation. First order interpolation or linear order interpolation the erroneous sample is replaced by a sample that is derived from the mean of the previous and subsequent samples. Combinations of zero crossing and first order interpolation are used in audio systems.
Muting is utilised as a last effort and sets the value of the uncorrected words to zero. Silence is preferable than letting the result of decoding incorrect data being heard [as clicking or popping]. To minimise the audibility of a mute, muting algorithms gradually attenuate the outputs signal amplitude prior to the mute and gradually restores the amplitude after the mute has been accomplished. [1]
+++++++++++++++++++++
Reference:
1. Principles of Digital Audio 5th Edition – Ken C. Pohlman
2. The Art Of Digital Audio – John Watkinson
3. Modern Recording Techniques DM Huber
4. Studio Recording Engineer
5. Microsoft Windows Help Menu
6. www.sweetwater.com |
|
Sun Oct 02, 2005 10:28 am |
|
|
|
|
Polarman
Joined: 24 Jun 2005
Posts: 55
Location: Barbados |
| Week 5 |
|
|
1. What is the sub-code area of the DAT tape used for?
Subcodes tell the DAT player the number of each selection or program, where each selection starts, and whether or not to play each one. Subcodes are independent of the audio signal written on tape. There are three main types of subcodes, Program Numbers, Start IDs, and Skip IDs. You can record or erase subcodes without affecting the audio program.
Program Numbers
Program Numbersare assigned to selections in order, and can then be used to locate them.
Start ID
A Start ID marks the beginning of each selection. It can be written manually or automatically. Manual Start IDs can be placed anywhere. Automatic Start IDs are put on tape whenever there are enough signals applied, after a silence of three or more seconds.
Like other subcodes, Start IDs can be recorded or erased without altering the audio program. You can enter them during recording or playback. If the cassette's safety tab is set to prevent accidental erasure, you can't record or erase subcodes. Most DAT machines can automatically re-number the Start IDs in consecutive order.
Skip ID
A Skip ID makes the machine skip the selection. This ID can only be written manually. Whenever the machine senses a Skip ID during playback, it stops and fast winds to the next Start ID, and begins playing. This function can be turned on and off.
The subcode is important in DAT player to be able to wanted areas on the tape can be reach rapidly and to be able hear some sound in shuttle so one can locate the right area.
http://www.tape.com/techinfo/datbook.html
Watkinson, J. (2001). The Art of Digital Audio, 3rd ed. Oxford, Focal Press.
2. Explain the principles of predictive coding.
There are two main categories of audio coding lossless and perceptive. The principles of predictive recording is a lossless method which means that the signal being played back is exactly as the signal recorded (minus errors ).
Audio signals are largely repetitive, which is why predictive coding works. The technique involves a 'predictor' which has knowledge of typical audio signal behavior. By looking at the preceding audio signal, it tries to anticipate what will happen next and, because of audio's repetitive nature, the prediction is generally quite accurate. If this prediction is subtracted from the original signal, only a small difference signal remains, and this is recorded or transmitted as the data-reduced result.
Both the coder and the decoder uses the same predictor 'knowledge' to generate/regenerate the predicted signal. The accuracy of this system is entirely dependent on the predictor algorithm. Ideally if the residual is transmitted intact there is no loss of information but in practice typically around 98% of the original signal is retrieved.
The technique works less well in anticipating essentially random signals in noise-like sounds, or in predicting highly unpredictable (but crucial) transients. To improve the precision of the system, therefore, many coders use band-splitting techniques (splitting the whole audio spectrum into four separate frequency bands). This allows multiple predictors to work on simpler band-limited signals with far greater accuracy than they would if handling the complete signal.
One drawback of predictive coding is that since the decoder must use exactly the same predictor as the encoder, improvements to the 'intelligence' of the encoder can only be useful if your decoder is updated too, otherwise the accuracy of the decoded signal will actually suffer.
In general, this kind of system works very well, providing a typical reduction ratio of around 4:1. However, it can prove fatiguing to the listener over long periods, because damaged transient signals require more 'brain power' from the listener to interpret the sound. Multiple passes through the encoding/decoding process also lead to rapid loss of signal quality, very similar to that experienced when copying analogue cassettes.
http://www.soundonsound.com/sos/aug98/articles/datacompression.html
Watkinson, J. (2001). The Art of Digital Audio, 3rd ed. Oxford, Focal Press.
3. How does the data buffer of a Minidisc player facilitate editing?
The Minidisc player has an in built data buffer and control of the data rate of playback. This was originally built in to allow the recovery from jogging the player and skipping data. It may also be used for editing, allowing pieces to be jumped, repeated or rearranged in order without any audible effect. Any discontinuity brought about by location is smoothed out by the data.
When you are editing on a minidisk its not the actual audio you are editing it is the TOC. Since the buffer holds enough data there will be no pause or clicks when skipping material.
http://www.minidisc.org/beyond_the_caddy.html
http://www.minidisc.org/faq_sec_5.html#_q40
http://radioworldwide.gospelcom.net/org/files/minidiscs.pdf
4. What are the basic principles of masking?
Frequency masking
Our brains do not treat the audio spectrum as a continuum the human hearing we perceive sound through around 25 distinct critical bands of varying bandwidths (at 100Hz the critical band is about 160Hz wide, but at 10kHz it is 2500Hz wide). You can say that humans listen through a 25 band EQ. Quiter sounds will be masked away from louder signals in the same band. This phenomenon is called frequency masking. Although our hearing is incredibly perceptive of simple signals in isolation, in the presence of complex sounds it effectively runs out of 'hearing resources' and so can only perceive the most dominant parts at any particular moment in time.
The hum from a bass guitar amplifier is inaudible while the guitar is playing, although quite evident on its own.Tape hiss is inaudible in the presence of full-range music, but obvious between tracks.
Time masking
A loud sound affects our perception of quieter signals both before and after it. A quiet signal that occurs 10-20 milliseconds before a louder one, for example, may be masked by the louder signal -- this is called backwards masking. The squeak of a kick-drum pedal might be plainly audible on its own, but can be masked by the presence of a much louder bass drum thump which happens a few milliseconds later. The hearing mechanism also takes time to recover from a loud sound, and this creates a masking effect which extends up to 100-200 milliseconds after the masking signal has ceased -- this is called forward masking. The length of the masking is related to the amplitude of the masking signal.
http://www.soundonsound.com/sos/aug98/articles/datacompression.html
5. Explain the following, with reference to a HD editing system
Disc access time
Disk access time is an average of the time between initiating a request and obtaining the first data character. It includes the command processing, the average seek time (moving the read/write head to the required track) and the average latency (rotation of disk to the required sector). This specification must be given as an average, because seek times and latency can vary depending on the current position of the head and platter”.
http://www.pcmag.com/
Editing on a HD system doesn’t involve an actual editing on the original audio files. Because of the random access on a disk edit points and preview and reviews are done almost instantaneously. During an edit the disk has to work much harder then a playback because it has to provide audio files from two different places. This can become critical is there are many close spaced edit or long cross fades. During a cross fade the data rate is twice then on normal playback.
The amount of work on the disk can be reduced by a large buffer memory. To reduce the amount of delay moving the pickup on the disk to a particular track a frequent defragmentation of the disk is important. It can also sometimes be necessary to move the original files close to each other to be able to do the edits.
Watkinson, J. (2001). The Art of Digital Audio, 3rd ed. Oxford, Focal Press.
EDL
Since the actual editing in a HD editing system is not done directly on the original recordings saved on disk a list of the edit points is saved in the Edit Decision List. Here is the series of locations of information that need to be accessed in a particular sequence producing a required running order. The advantage of this is the edit is not satisfactory only the EDL has to be modified not the actual original recording.
Watkinson, J. (2001). The Art of Digital Audio, 3rd ed. Oxford, Focal Press.
Disc data bandwidth
The Disc data rate is the speed, normally in bits or bytes per second, which data can be written or read from the disk. When editing on a multitrack the bandwith gets crucial.
6. Explain briefly the problems of transferring data files between different proprietors HD systems.
Each manufacturer has different way of saving a project. The actual data or recording is relatively universal but the way to store cross fades, edits, dsp functions is different. One way to resolve this is to save each track as a wav file making it possible to import the tracks. The downsiade of this is that all edits all lost. One standard that is getting common is OMF (Open Media Framework) that makes it possible to transfer project between different HD systems.
7. Explain the principles used in editing on a single Mini Disc system.
The Mini Disk has a relatively fast access time and is fitted with a buffer delay memory. This allows audibly transparent jumps from one section to the other, creating an Edit decision list system at low cost
8. Describe and explain the following error handling systems
Causes of errors
There are lots of reasons why errors occur. Manufacturers can control noise, crosstalk and other potential interference so that, for all intents and purposes, there will be no errors.
Magnetic recording small random errors can affect single bits, and isolated large bursts of errors can disrupt a whole array of bits in an area that is otherwise error-free. Errors can be caused by Gaussian thermal noise in heads and replay circuits, or losses of head-to-tape contact resulting from imperfections in the magnetic coating, small bits of dust from the media itself and/or improper storage and handling.
Contamination in optical recording can interrupt the light beam. Warped disk can cause defocusing.
Even an environment as seemingly safe as the inside of a memory chip can have problems. The tiny wells of capacitive charge that represent zeros and ones can be discharged by alpha particles from the natural radioactive decay of the chip's own materials. Statistically this is only going to happen once every 30 years or so, but with thousands of chips in a large memory bank, the probability rises to an error every few minutes. In cables electromagnetic interference and crosstalk can occur and corrupt data.
Every digital channel has its own set of problems, and the solutions applied will be different for each type of channel.
Minimisation of the consequences of errors
The first steps in error management constitute a sort of pre-processing in anticipation of the errors to come. Much of this is simply good engineering, doing as much as we can to avoid errors. We design circuitry to minimize noise and crosstalk. We find bad spots on hard discs and lock them out. We see to it that there is enough transmit power and a good enough antenna to ensure an adequate signal-to-noise ratio at the receiver.
Error detection
To detect errors redundancy coding is used. Without it, error detection would be impossible. Detection is one of the most important steps in error management. It must be very reliable. If you don't know there has been an error, it doesn't matter how effective your other error management techniques are.
Redundancy codes can be extremely complex but one simple way is to use parity check. Like all redundancy codes, the parity check adds bits to the original data in such a way that errors can be recognized at the receiver. There are not more bits recorded only used for error detection. A parity bit is added to the data words. By adding a zero or a one, the parity bit ensures that there will be an even number of ones in all the coded data words. If the error detection circuitry in the receiver sees an odd number of ones, it knows an error has taken place.
Irrespective of the cause of the error there are two main effects. Larger isolated corruptions called bursts which occur in an area which is normally error free and there is random error which affects single bits or symbols.
Error correction
In all error correction adding some bits calculated from the data is used. The new entity is called codeword. Codewords are design to be able to correct totally a finite number of corrupted bits. Once an error correction system is used the signal to noise ratio can of the channel can be reduced. That why error correction systems doesn’t reduce storage or transmission capacity. Its also important to mention that once an error is corrected there is no audible difference from the original.
Error concealment
Errors will occur that cannot be corrected. The only option is to conceal them as best we can.
With digital audio, the simple fix is to approximate a lost sample by interpolating a value from samples on either side. A more advanced method makes a spectral analysis of the sound and inserts samples with the same spectral characteristics. If there are too many errors to conceal, then the only choice is to mute.
Prior to recording the sample stream the odd and even samples are separated. They may even be recorded in separate places to avoid that a burst error affects both the odd and the even samples. On replay the odd and even samples are put together as the original sequence. If an error has occurred you will at least have every other sample correct. The incorrect samples are recalculated. If sample 7 and 9 are correct and sample 8 is incorrect you can calculate sample 8 as (7+9)/2.
Watkinson, J. (2001). The Art of Digital Audio, 3rd ed. Oxford, Focal Press |
|
Tue Oct 04, 2005 12:38 pm |
|
|
|
|
All times are GMT. The time now is Fri May 16, 2008 4:04 pm
|
|
|
|
| |
