WO2009118252A1

WO2009118252A1 - Method and apparatus for storing data to a storage medium, method and apparatus for correcting errors occurring while data is read from a storage medium, and storage medium

Info

Publication number: WO2009118252A1
Application number: PCT/EP2009/053012
Authority: WO
Inventors: Marco Winter
Original assignee: Thomson Licensing
Priority date: 2008-03-27
Filing date: 2009-03-13
Publication date: 2009-10-01
Also published as: EP2105928A1

Abstract

AV content stored on prerecorded storage media often does not entirely consume the available media capacity, and an unused special area will remain. Also, known recording formats often provide areas which are free from any information that is relevant for reproduction and, therefore, represent a further special area. The invention proposes to derive safety information from the data of the known recording format, to embed the safety information into the special areas, to store it along with the data, and to use the safety information for restoration of the read data in the event errors are occurring. Preferably, the safety information of a specific section is stored to special areas of a previous section, thus allowing real-time correction.

Description

Method and apparatus for storing data to a storage medium, method and apparatus for correcting errors occurring while data is read from a storage medium, and storage medium

The invention relates to a method and an apparatus for storing data to a storage medium according to Claims 1 and 6, to a method and an apparatus for correcting errors occurring while data stored on a storage medium is read according to Claims 7 and 9, and to a storage medium according to Claim 11.

The methods are particularly intended for the manufacture of so-called masters, i.e. casting molds for, e.g., CDs, DVDs, HD-DVDs, and BDs, as well as for DVD recorders which are, for example, used in private applications for playing back and/or storing television programs and videos. The storage media are, for example, masters, CDs and DVDs.

If video and/or audio data is stored to a storage medium in the form of digital data records - a process that is also referred to as writing to a storage medium - then this is achieved in the form of data records in a plurality of sectors according to a predetermined convention which is also called "format" or "standard". In the following, the data records will also be referred to as data, to simplify matters. In most cases, the storage capacity of the storage medium is, herein, not fully utilized. That means that there is still free storage space left on the storage medium.

While the storage medium is read - a process which is also referred to as playing back - one or more sectors may occa- sionally be corrupted in an irreparable manner, despite an error protection contained in the predetermined convention. Where audio data records are concerned, the corruption can be compensated or masked to a certain limit by means of an interpolation software. If the limit of what can be masked is exceeded, there will be dropouts and/or "skips". Where video data records are concerned, however, it is presently not possible to achieve unnoticeable compensation, and there will be brief freeze frames or what is called block artefacts during playback.

The invention aims at creating a method for the storage of data to a storage medium wherein the data is stored in such a way that any corruption of data records possibly occurring at a later time can be repaired while the storage medium is played back.

A further object of the invention is to create a method for the correction of errors occurring while data stored on a storage medium is read in a player, which method allows an error correction of the data stored on the storage medium if necessary.

Further objects are the provision of apparatuses which can be used to apply the methods according to the invention.

A further object is the provision of a storage medium which contains such a kind of data that allows fast and easy correction of erroneous data by means of the methods according to the invention.

The first object is achieved by the elements of Claim 1. Safety data records are formed in addition to first data which, as a whole, represent a set of data which is to be stored according to a predetermined convention and which can be reproduced according to the same convention, wherein said first data contains, at irregular positions, special areas which are not used during reading according to the predeter- mined convention. The first data records and the safety data records are jointly stored to the storage medium. The safety data records each contain safety information about parts of the first data records, said safety information being of such a kind that it allows recovery of erroneous first data re- cords while itself requiring only relatively little storage space .

In other words, for storing data to a sector-organized storage medium, the data is stored to first address ranges of the storage medium, said first address ranges being predetermined according to a convention; safety data records are formed which contain safety information about the data; the data is used to derive positions of second address ranges of the storage medium, said second address ranges not being aligned at sector boundaries; and the safety data records are stored to the second address ranges of the storage medium. These steps are not necessarily performed in the order specified.

According to known storage methods, the data is partitioned into sections according to the predetermined convention and are thus stored to the storage medium. For example, an often used form of section for video data is the VOBU = Video Object Unit corresponding to a playback time of about 0.5 sec- onds . According to the DVD Video convention, VOBUs can vary both with regard to their playback time and with regard to their data volume, even within one and the same recording. The data of one section will be referred to as a data section below. Irrespective of the partitioning into sections, it is wide-spread practice to store the data to sectors of the storage medium which are constant in size, which medium is then called a sector-organized storage medium; and often, an additional agreement applies, specifying that each of the sectors contains data of one specific data type only. Data types are, among others, audio data, video data, subtitle data, or navigation data. Where video source encoding according to the MPEG-2 standard is concerned, video data is fur- ther subdivided into what is called "I-picture data" for single frames which can be decoded without referencing other video data of the frame sequence, "P-picture data" the decoding of which requires a reference to exactly one frame of the frame sequence having been transmitted previously in the data stream, and "B-picture data" the decoding of which requires a reference to exactly two frames of the frame sequence having been transmitted previously in the data stream.

According to the predetermined convention, for example the DVD video standard, there are, in most cases, still unas- signed storage locations in the data formatted for storage to the storage medium, both at the end of this data and within some of the sectors used. This comprises an unassigned storage location at the end of the medium whenever the total vol- ume of the data to be recorded does not exactly correspond to the storage capacity of the storage medium. It also comprises those cases where, according to the predetermined convention, parts of sectors or certain complete sectors remain unrecorded to form reserved areas or are filled with pseudo data, which are often also referred to as "dummy data" or "stuffing data". In the following, both types of areas together will be referred to as "special areas" and as second address ranges which are not aligned at sector boundaries. A common characteristic of the special areas or second address ranges is that, according to the predetermined convention, the data contained therein is not used or required for reading the first data.

The invention utilizes the special areas by storing the safety data records exactly to these areas. Advantage is taken of the fact that, during readout according to the predetermined convention, the special areas are skipped from the outset or are not considered at all. In other words, the content of the special areas is irrelevant for reading out according to the predetermined convention. This ensures that the process of playing back according to the convention in a prior art player is not affected negatively, even if data other than that prescribed according to the predetermined convention is stored to these special areas. That means that all of the methods according to the invention are downward compatible with the predetermined convention. In other words, a storage medium which, according to the invention, contains safety data records embedded in the first data records, can be played back on a prior art player according to the predetermined convention, as well - then, however, without the advantages of the additional error protection contained in the safety data records.

By accommodating safety data records in the special areas, the safety information which can be used to repair possibly corrupted first data records will subsequently be available when the data carrier is being read in an apparatus according to the invention. While the data stored according to the invention is played back by a player according to the invention, not only the first data records are read out according to the convention but also the safety data records are read out and are buffer stored if necessary. If an erroneous first data record occurs, the player according to the invention makes a correction by logically combining the first data records and the safety data records, and achieves an error-free reproduction whenever there are not too many erroneous first data records. If played back by a player according to the invention, a data carrier showing signs of usage of a certain intensity and also comprising safety data records according to the invention does not generate any reproduction failures, not even if failures are already occurring when played back by a prior art player.

The method according to the invention can be used for the storage of data to various storage media. For example, it can be used during the manufacture of a so-called master, that means of an intermediate product which, in turn, is used for mass production of a "pre-recorded data carrier" as finished product. The master, and with it all finished products manufactured therewith, will also contain the safety data records in addition to the first data records according to the convention. Further possibilities of application exist, for example, during "burning" and/or copying of CDs/DVDs. Since the method does not utilize any special features of optical data storage, it can also be used for storage media according to other physical principles, e.g. for magnetic storage media.

Preferably, the safety data records are arranged on the storage medium in such a way that the safety information contained therein can be used to allow a real-time error correc- tion of the first data records. If the data is subdivided into a sequence of sections for consecutive reproduction, this is achieved if the safety data records relating to the first data records of a specific section are stored embedded in the first data of at least one of the previous sections. In the case of video data, a section can for example be agreed upon to equal a VOBU; with the result that the safety data records of a VOBU no. (n) are stored along with the first data of the previous VOBUs no. (n-1) and/or no. (n-2) . In this manner, the safety information contained in the safety data records and pertaining to a section are already completely read out during playback, if a possibly defective first data record is detected among the first data records of the section. This allows very quick recovery of the defective data. If the safety data records pertaining to a section would be embedded after the first data records of the section, a jump back on the storage medium would be required every time a defective first data record is detected.

In a further embodiment of the invention, the safety information comprises parity data which is formed from the first data records. The parity data is logically exclusive-OR combined data also referred to as EXOR data, which is generated from the first data records at low complexity in terms of implementation. The data is buffer stored and then stored to the special areas on the storage medium. Instead of parity data derived from sections, it is also possible to use other error protection data according to known error correcting en- coding methods, e.g. from block codes or from convolutional codes .

It is also possible to form and store two safety data records for each section, wherein one is formed across odd-numbered first data records and the other is formed across even- numbered first data records of the section. This allows repair of data, i.e. correction of errors, even in those cases where, within one section, two defective first data records exist, spaced apart by an odd number of records, which includes the most important special case of two directly neighboring defective first data records. The cost for this benefit is that, at an otherwise constant context, the volume of the safety data records is doubled. Generalizations toward three and more safety data records per section are also known from the field of RAID technology.

According to a further variant of the invention, advantage is taken of the fact that audio-visual data generally contain data that is required, essential or important for playback and supplementary data that is less important. The variant consists in that the safety information to be accommodated in the safety data records is exclusively formed from parts of the first data records that are declared to be essential. Assuming that one safety data record is to be provided for a constant number of first data records, the limitation to the essential first data records reduces the volume of safety data records to be stored additionally and, therefore, the free storage space required. Alternatively: if one assumes a constant number of safety data records per section, then the condition for the invention's methods to work, namely that exactly one of the first data records associated to a protection data record is defect, occurs at an error probability that is the greater the lower the number of first data records per safety data record is. Depending on the type and volume of the first data records selected to be essential, protecting them by means of associated safety data records enables that, if the protectable case occurs, data are repro- duced either without any visible or audible errors at all, or with just minor errors. The approach of subdividing the first data records into essential and unessential data records and, by agreement, forming the safety data records only from the essential data of the first data records, can for instance be implemented in that the safety information is formed from only a part of the data types I-picture data, P-picture data, B-picture data, and non-video data. Among these, the I-picture data has to be considered to be the most essential, because it is required and used both for its own decoding as well as for the decoding of the P-pictures and the B-pictures relating thereto. That means that, if the data stream available for error protection according to the invention is limited, in general I- picture data will be protected with top priority.

The method according to the invention is somewhat similar to the "RAID 5" method known from computer technology. According to that method, parity data is formed from the data to be stored and is uniformly stored to a plurality of hard disks according to a predefined system.

An apparatus according to the invention for the storage of data according to the elements of Claim 6 allows application of the storage method according to the invention.

The object of creating a correction method is solved by the elements of Claim 7. The method according to the invention allows to correct erroneous first data records while data are being read. In most cases, this can be achieved in real time. The method allows error-free playback of, e.g., DVDs even if some first data records are corrupted. The errors are, there- fore, not noticed by a user.

In a variant according to Claim 8, the corrected first data records are stored to a recordable non-volatile data storage of the player. This facilitates error-free playback and allows error-free playback of the non-recordable storage medium at a later point even if mechanical wear and tear causes the latter, for example, to become so poor over time that, some- time, it is no longer possible to correct the erroneous first data records exclusively from the data of the medium.

It is advantageous if the safety information stored to the recordable non-volatile data storage of the player addition- ally comprises an identification code of the storage section to which it pertains. Such an identification code consists, by way of example, of an identification code of the storage medium and a section identification code. The existence of a storage section identification code allows that the safety information can directly be associated to a specific non- recordable storage medium and to the identified erroneous first data records on it. This facilitates and accelerates error-free playback at a later point.

In a further embodiment of the invention, what is stored to the recordable data storage of the player are not only corrected erroneous first data records but also, along with the latter, those first data records that neighbor such corrected erroneous first data records. This takes into consideration that there is a high probability that these neighboring first data records become erroneous within a relatively short time, e.g. in case of a slightly increased mechanical wear and tear of the storage medium, and therefore obviates potential un- readability .

In a further embodiment of the invention, a prompt can be output to the user of the player requesting him or her to backup the optical storage medium as soon as the latter comprises a certain minimum number of errors. As a result, the user can make a copy of the optical storage medium before it becomes unreadable as a whole or in parts.

An apparatus according to the invention for the correction of errors according to the elements of Claim 9 allows application of the correction method according to the invention.

The object of providing a storage medium is solved by the elements of Claim 11. The storage medium contains safety data records which contain safety information about the first data. Herein, the safety data records are stored to the second address ranges on the storage medium, that means to those locations which would remain unassigned according to the pre- determined convention or the content of which is not required for reading out according to the predetermined convention. This ensures that the storage medium can be played back without any errors even if some first data records comprise errors, for example caused by scratches. If the storage medium is a master, said master can be used, for example, to mass- produce corresponding media for end users, said media containing all data of the master and, therefore, the safety data records as well.

The independent claims are linked by the single general inventive concept that from first data records which are required for reading data according to a convention, safety data records are derived which contain safety information about the first data records, and that these safety data re- cords are stored to storage locations or address ranges which remain unassigned according to the convention. The safety information allows correction of errors potentially occurring in the first data records, with the result that the data can still be output without any errors.

The invention will be illustrated in more detail by means of the examples presented in the drawings and the dependent claims. In the drawings,

Figure 1 shows an exemplary curve of a data stream;

Figure 2 shows a formation of parity data;

Figure 3 shows a storage of a parity block;

Figure 4 shows a storage of a parity block formed from clus- ters;

Figure 5 shows a storage of parity blocks formed from clusters, said storage being an alternative of the storage shown in Figure 4; and

Figure 6 shows a block diagram of a player according to the invention.

An apparatus for storing data to a sector-organized storage medium comprises at least one internal recordable data storage. In addition, means for the formation of safety data re- cords from the data, means for buffer storage of the safety data records to the internal data storage, means for deriving, from the data, positions of second address ranges of the storage medium which are not aligned at sector boundaries, and means for storage of the safety data records to the sec- ond address ranges of the storage medium are arranged.

In commercial applications, the apparatus is, for example, a device for the manufacture of masters or, in private applications, a recorder for optical storage media. Accordingly, the storage medium is, for example, a master disk or a recordable or re-recordable CD or DVD or other storage medium. To be stored, data containing at least one digitized audio/ video data stream, which is also referred to as AV data stream, or an audio data stream is transmitted to the apparatus provided for storage. This data comprises first data records which are required for reading the data according to the predetermined convention.

The data is continuously buffer stored to the internal data storage. According to the predetermined convention, the data is subdivided into sections and the sections are subdivided further into sectors, unless the data is already provided in subdivided form when it is fed into the apparatus for storage. Where AV data is concerned, a section, for example, corresponds to a VOBU. According to the convention, each sector only contains data of a specific data type, e.g. audio, video, subtitle, or navigation data, and comprises a pack header consisting of MPEG2 data. That means that, if, in a section, a data type comprises only few data or does not exist at all, at least parts of the corresponding sectors remain more or less unassigned according to the predetermined convention. Furthermore, the set of data to be stored to the storage medium corresponds to the storage capacity of the storage medium very rarely only. According to the convention, there will, therefore, almost always remain free storage space on the storage medium even after completed storage.

While the data is processed, safety data records are additionally formed according to the invention. These safety data records contain safety information about the first data records. The safety information allows correction of the first data records if errors are occurring there.

For example, the safety data records comprise parity data which is calculated from specific sector sequences of the first data records. The parity data is then combined in EXOR sectors to form parity blocks. This is achieved in a manner similar to that used in the RAID 5 system when a plurality of hard disks arranged in a computer are secured and managed to- gether. Each byte of an EXOR sector results from the exclu- sive-OR combination of bytes of equal order pertaining to a plurality of sectors. That means that an x-th byte of an EXOR sector is obtained by exclusive-0R combining the respective x-th bytes of the allocated sectors. Preferably, allocation data, for example specifying the address and the length of the VOBU from which the EXOR sector is derived, are also added to the EXOR sector or to those of its parts that are to be stored. Further EXOR sectors can, for example, be formed for other video sectors and/or only for the I-picture data of a video data stream encoded according to MPEG. The latter represent a subset of video sectors which are particularly essential for playback.

Advantageously, the parity data is exclusively formed from essential data types. These include, for example, video data, I-picture data, P-picture data, and B-picture data. Navigation and subtitle data, however, are of less relevance and can therefore be omitted.

Even within sectors containing essential data types, only parts may, in fact, be relevant for reproduction; other parts may be able to be restored relatively easily. For example, data that can be restored easily are the pack header and the packet header which are found in each sector according to MPEG. That is why this data can also be omitted from the formation of parity data. This is advantageous in that it reduces the volume of the parity data. Additional data integrity is achieved by deriving two first data records from the sectors of a section, where one of these first data records is derived from odd-numbered sectors and the other one from even-numbered sectors. This allows restoration of two neighboring unreadable sectors in the section .

The formation of the safety data records can be achieved simultaneously with the storage of the first data records to the storage medium.

The safety data records are buffer stored to the internal storage medium. From the internal storage medium, the safety data records are stored to the storage medium. This is achieved at the second address ranges, i.e. at those storage locations of the storage medium which remain unassigned or are only filled with pseudo data, stuffing data or dummy data when the first data records are stored according to the predetermined convention. Herein, the safety data records of a VOBU no. (n) are preferably stored to the second address ranges of the VOBUs no. (n-1) and/or no. (n-2) preceding no. (n) in readout order.

In cases where the additional error protection according to the invention is to be applied in a particularly strong manner, it is possible - even without violating the predetermined convention regarding the format of the first data records - to generate additional free space by distributing the payload of rather rarely occurring packets, e.g. sub-picture packets, across more sectors than would be necessary due to their data volume. If the storage capacity of the unassigned storage locations in the VOBU no. (n-1) or (n-2) does not suffice for all safety data records of the VOBU no. (n) , a selection according to importance takes place and the most important safety data records are stored first. The remaining safety data re- cords are appended at the end of the data stream if sufficiently many free locations are available there. Such a procedure allows real-time restoration in case of data errors. At the end of the data stream, too, safety data records can be stored according to their importance, if the storage space does not suffice for all safety data records.

At least one storage medium is manufactured in this manner. As a result, safety information allowing error correction during subsequent playback is available on the storage me- dium. The processes are described in more detail by means of Figures 1 to 5.

As described above, safety data records are derived from the first data records, so that the total volume of the data to be stored results from the sum of the volumes of the two types of data records. Figure 1 shows an example of data streams R which may occur over time t if the method according to the invention is applied. An upper horizontal line 100 indicates that hypothetical constant data rate which can maxi- mally be continuously read from the storage medium. An upper variable line 101 indicates the time-dependent behavior of the data stream which results, according to the predetermined convention, for a hypothetical example AV material. As has been described above, this data stream also contains special areas. A middle variable line 102 indicates the time- dependent behavior of the data stream without these special areas, i.e. the time-dependent volume of the data actually required for reproduction according to the predetermined convention. The vertical distance between the upper variable line 101 and the middle variable line 102 thus indicates, for each time t, the momentary volume of the special areas described in more detail above; hence it represents the bit rate that is directly available for the purposes of the method according to the invention. By additional modifications of the time stamps of the sectors or by other modifications of the encoding method according to the predetermined convention, the maximum total bit stream that can be used for the purposes of the method according to the invention has a size which results from the positive distance between the upper horizontal line 100 and the middle variable line 102. In Fig. 1, this bit rate is also shown separately or non- incrementally as the broken line 103. Fig. 1 also shows, as a lower horizontal line 104, an average safety bit rate which can be arbitrarily chosen within certain limits. For the example shown in Figure 1, a specific value is assumed which is much smaller than the total bit rates represented by the upper variable line 101 or by the middle variable line 102.

A lower bold variable line 107 in Fig. 1 indicates the time- dependent behavior of the size of the data stream that is actually used for the error protection according to the invention. It can be seen that, at all times t where the size of the available special areas allows, the safety data records are generated and stored at the average safety bit rate.

Since the constant data rate 100 must not be exceeded, and cannot be exceeded on some players, it is at those time posi- tions of the data stream where the constant data rate 100 is reached, that the associated safety data records are stored to previous VOBUs which still provide enough free storage ca- pacity. In doing so, the safety data records relating to the first data records of a VOBU no. (n) are stored to the special areas of the preceding VOBU no. (n-1) and, if necessary, to further previous VOBUs (n-2) etc.. It is additionally taken into account that safety data records of any specific VOBU must always be stored before safety data records of a subsequent VOBU. In this, the distance between a VOBU and the storage location of the associated safety data records should be made as small as possible, thus ensuring fast availability of the safety data records in the event of a necessary re- pair.

The safety data records can be buffered best with an offline encoding because a real-time encoding requires essentially more available buffer storage space.

Figure 2 illustrates the formation of the safety data records in the form of parity data. Herein, only the essential data types are, preferably, used to form parity data, which are then combined to form a parity block 213. In its upper part, Fig. 2 shows several sectors 209 of consecutive VOBUs 201,

..., 208, wherein the data that is essential for parity formation is marked by right-diagonal hatching and the unessential data is shown without any hatching. In its lower part, Fig. 2 shows in more detail the formation of a parity block 213 from those sectors of three successive VOBUs 204, 205, 206, that contain essential data. In the figure, the parity block 213 is marked by left-diagonal hatching.

In particular, in the central region of its lower part, Fig. 2 shows that, where sectors contain both essential and unessential data, the essential data is included in the parity formation, whereas the unessential data, the position of which is known, is ignored or omitted in the parity formation. Equally indexed bytes of all essential data areas 210 of the involved sectors are logically combined, as is indicated in Fig. 2 by the arrows from left to right, each yielding one byte of the parity block 213. The number of VOBUs across whose sectors one parity block 213 is formed, with other words the size of each section, can also be varied dynamically, depending on the unassigned storage locations available in the particular section. This will be described in more detail below.

Figure 3 shows by way of example how a parity block 213 that has been derived from the essential data of a section 211 is stored. To this end, the parity block 213, again marked by left-diagonal hatching, is subdivided into suitably sized parts 220, 221, 222, 223, 224, and each of these parts, along with a preceding header 214, is stored, that means filed, to one of the special areas of the sectors of previous VOBUs 201, 202. Since the size and the position of the special areas are known when the data stream is being decoded according to the predetermined convention, the size of the parts into which the parity block 213 must be partitioned is also defined and does not have to be separately stored, coded or marked.

The example shown in Fig. 3 is based on the assumption that every parity block 213 is formed from three consecutive VOBUs, in the manner described. Alternatively, parity blocks 213 can also be derived without taking VOBU boundaries into account, either from a dynamically varying number of sectors or from a constant number of sectors. One way to achieve this is to partition the data stream into sector clusters, selecting the size of the sector cluster in such a way that each sector cluster comprises just as much free storage space in its special areas as is necessary for storing the parity block of the subsequent sector cluster.

Figure 4 symbolically shows a data stream whose sectors per- tain to consecutive VOBUs 201, ..., 208 and have been grouped into variably sized sector clusters 211 depending on the size of the special areas contained therein. The special areas of each sector cluster contain the parity block formed across the essential data of the subsequent sector cluster. For the sake of clarity, in Fig. 4 parts of parity blocks are shown with vertical hatching whereas the unused parts of the special areas are left without any hatching. It is, in particular, apparent from Fig. 4, as for instance in the next to last sector of the VOBU 207, that using this data arrange- ment, that special area within a sector cluster into which the end of the parity block is stored, is only partially filled; the end of the special area remaining empty.

Figure 5 shows an alternative data arrangement which avoids this drawback, uses the free storage space even better, and thus ensures improved correction of defective data records. Again, it is assumed that that sector is considered to be the end of each sector cluster 211, whose special area contains the end of the parity block of the subsequent sector cluster 211. In addition, it is now assumed that, if the end of the parity block that is stored in the last special area of the sector cluster 211 fails to fill the special area completely, the remaining free storage locations of that special area are used to receive at least the beginning of the parity block of the next sector cluster 211. To better illustrate these rules in Fig. 5, the parts of the parity blocks are alternately marked by vertical and left-diagonal hatching. At a sector cluster boundary it is therefore in this case possible to store parts of successive parity blocks in one sector. On the other hand, it is not mandatory to use all unassigned storage locations of the special areas, which are available as such. It can for instance be advantageous to ensure that every sin- gle parity block part has at least a specific minimum size.

In one embodiment of the invention, parity blocks are subdivided into parity block parts according to the size of the special areas available in those sectors into which they are to be stored. For the purpose of allocation, each of the parity block parts is provided with a header 214 which contains, for example, details on the size and, if necessary, the beginning of the parity block, on an offset of the parity block part within the parity block, on a cyclic redundancy check (CRC) or a checksum of the parity block part, and/or a magic word. Each bit pattern which appears as rarely as possible or never in the regular bit stream of the first data records can be used as magic word. Through embedding a magic word, it is possible to identify the parity data in the total data stream in a reliable and/or fast manner.

As an alternative to providing each parity block part with a header, only the first parity block part of each parity block may have such a header or a similar header; or the headers of subsequent parity block parts of the parity block may comprise fewer details.

Independent of whether sector clusters or VOBUs are used as sections from which the parity blocks are derived, it is pos- sible to derive and store, as described above, more than one parity block per section. In this manner, repairing a plurality of consecutive defective sectors becomes possible. It is furthermore advantageous if a plurality of safety data records are provided and jointly embedded into the first data records, if these safety data records are formed from differently sized subsets of the first data records. To achieve this, it must be taken into consideration that, if parity data are used as safety data, the methods according to the invention provide their error protection effect if one and only one of the first data records each pertaining to a safety data record is erroneous. If, among the first data re- cords that are logically combined into a parity data record, two or more are erroneous, correction by means of parity data is no longer possible.

Let us now consider a storage medium that is gradually becom- ing poorer and poorer through increasing mechanical wear and tear, i.e. a storage medium that exhibits a slowly growing read error probability. As long as, initially, the wear and tear is very low, a basic error protection that is presumed to exist according to the predetermined convention will be able to correct all read errors, with the result that the methods according to the invention are not being activated. If the error probability gets somewhat higher, the "protect- able situation", i.e. the case that no more than one first data record per safety data record is erroneous, will tend to occur first for those safety data records that were derived from the biggest number of first data records. If the error probability grows even further, an increasing number of double or even triple errors within the allocated first data records will occur for this type of safety data records. With an increasingly higher error probability, the single errors required for application of the methods according to the invention will increasingly occur in those other types of safety data records that are based on smaller numbers of first data records.

An appropriate decoding method for this case, therefore, consists in examining, in order of decreasing cardinality, the sets of first data records, which pertain to the safety data records of each section, with regard to whether exactly one first data record is erroneous in the particular set concerned. As soon as this situation is detected, the single erroneous first data record is corrected by means of the safety data record; and for reproduction of the section, a reproduction method must be used which is compatible with the scope of error-free data contained in the current set of first data records. If, for example, a storage medium contains first safety data records which are formed from all video data re- cords, second safety data records which are formed from I- and P-picture data alone, and third safety data records which are formed just from the I-picture data; and if, in addition, the protectable situation arises for a second safety data record of an area, then a reproduction method only requiring I- and P-picture data can be selected for this area. Such a reproduction method would, for example, consist in that, instead of decoding and reproducing B-pictures, the respectively previously decoded I- or P-pictures are repeated suitably often.

Figure 6 shows a block diagram of a player comprising means for the correction of errors occurring while data stored on a storage medium is read. The player comprises means for the detection of errors 5, means for the evaluation of the safety data records having a filter 4, a first buffer memory 2, a second buffer memory 3, as well as means for the replacement 6 of the erroneous first data records with corrected first data records. The player is, preferably, combined in one device along with the apparatus for storage, e.g. in a video recorder. While the storage medium is played back, a laser 1 reads the data from the storage medium and buffer stores said data to the buffer memories 2, 3, wherein the filter 4 supplies the first data records to the first buffer memory 2 and the safety data records to the second buffer memory 3. During normal fault-free operation, the first data records are transferred to an output 10 via a demultiplexer 7, three decoders 8 and one mixer 9, according to the broken line.

As soon as the means for the detection of errors 5 detect an error in one of the first data records, the erroneous data record, along with the associated safety data record, is sup- plied to the means for correction 6 and is corrected, i.e. recovered, by these means with the help of the safety data record. The corrected first data record will then be supplied to the demultiplexer 7 at the right moment and, thus, will be integrated into the data stream at the proper location and processed further. Where audio data is concerned, the mixer 9 is bypassed.

According to a particular embodiment, the player additionally comprises an internal non-volatile data storage and means for the storage of the corrected first data records on the internal data storage. If, in this case, an error is detected while a non-recordable optical storage medium is played back, the corrected first data record is stored into the internal data storage along with a section identification code and an optical storage medium identification code. At the same time, it is also possible to store into the internal data storage sections that are physically close to an erroneous section , because the probability that these neighboring sections will also become unreadable is high. If an optical storage medium has exceeded a predefined limit value of errors, a signal may be output to the user of the player, recommending backup copying the optical storage medium. During such a backup copying process, erroneous sections will then be replaced with the corrected sections.

In other words, AV content stored on prerecorded storage media often does not entirely consume the available media ca- pacity, and an unused special area will remain. Also, known recording formats often provide areas which are free from any information that is relevant for reproduction and, therefore, represent a further special area. The invention proposes to derive safety information from the data of the known re- cording format, to embed the safety information into the special areas, to store it along with the data, and to use the safety information for restoration of the read data in the event errors are occurring. Preferably, the safety information of a specific section is stored to special areas of a previous section, thus allowing real-time correction.

Claims

1. A method for storing data to a sector-organized storage medium, comprising the steps of:

- storing the data to first address ranges of the storage medium, said first address ranges being predetermined according to a convention,

- forming safety data records containing safety information about the data, the method characterized by the steps of: - deriving, from the data, positions of second address ranges of the storage medium which are not aligned at sector boundaries,

- storing the safety data records to the second address ranges of the storage medium.

2. A method according to Claim 1, applied to data which is subdivided into a sequence of data sections to be successively reproduced, wherein the safety data records formed across one of the data sections are stored to those sec- ond address ranges of the storage medium that are arranged within a previous one of the data sections.

3. A method according to one of Claims 1 to 2, comprising the step of selecting essential data from the data, wherein safety data records are exclusively derived from the essential data.

4. A method according to Claim 3, wherein the data comprises at least two data types from a set of data types, and wherein the essential data is selected by unselecting the data of at least one of the at least two data types.

5. A method according to Claim 4, wherein the set of data types comprises an I-picture data type, a P-picture data type, a B-picture data type, and a non-video data type.

6. An apparatus for storing data to a sector-organized stor- age medium according to the method according to any of

Claims 1 to 5, the apparatus comprising:

- at least one internal recordable data storage,

- means for forming safety data records across the data,

- means for buffer storage of the safety data records to the internal data storage,

- means for deriving, from the data, positions of second address ranges of the storage medium which are not aligned at sector boundaries, and

- means for storing the safety data records to the second address ranges.

7. A method for correcting errors occurring while data stored on a sector-organized storage medium is read in a playback apparatus, comprising the steps of: - deriving positions of second address ranges of the storage medium which are not aligned at sector boundaries from data read from first address ranges of the storage medium which are predetermined according to a convention,

- reading safety data from the second address ranges, - buffer storing the safety data,

- if data from the first address ranges is erroneous, using the safety data for the correction of the erroneous data .

8. A method according to Claim 7, wherein the corrected data is additionally stored to an internal data storage of the playback apparatus.

9. A playback apparatus having means for correction of errors occurring while data stored on a sector-organized storage medium is read, the apparatus comprising:

- means for reading first data from first address ranges of the storage medium which are predetermined according to a convention,

- means for determining, from the first data, positions of second address ranges of the storage medium, which are not aligned at sector boundaries, and - means for reading, from the second address ranges, safety data sets containing safety information about the first data, wherein the means for correction comprise means for the detection of errors, means for the evaluation of the safety data records, as well as means for the replacement of the erroneous first data with corrected first data.

10. An apparatus according to Claim 9, additionally comprising means for storing the corrected first data to an in- ternal data storage of the apparatus.

11. A sector-organized storage medium on which data is stored according to a predetermined convention, wherein the storage medium comprises: - first data in first address ranges predetermined according to the predetermined convention, characterized in that it contains, in non-sector-aligned second address ranges the positions of which can be determined from the first data and the content of which is not used for reading out according to the predetermined convention, safety data records containing safety information about the first data.

12. A storage medium according to Claim 11, wherein the data is subdivided into a sequence of data sections which are to be played back successively and wherein the safety data records formed across the first data of one of the data sections are stored to those second address ranges that are arranged within a previous one of the data sections .

13. A storage medium according to any of Claims 11 to 12, wherein the data comprises at least two data types of the I-picture data, P-picture data, B-picture data, and non- video data types, and wherein the safety information is not formed across all of these data types.