Error detection/correction and fault detection/recovery – Pulse or data error handling – Digital data error correction
Reexamination Certificate
1999-06-23
2002-07-16
Baker, Stephen M. (Department: 2784)
Error detection/correction and fault detection/recovery
Pulse or data error handling
Digital data error correction
Reexamination Certificate
active
06421802
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to audio coding methods and in particular to error concealment techniques in the coding of audio signals.
BACKGROUND OF THE INVENTION
The so-called “Internet Radio Broadcasting”, i.e. the transmission of a radio programme live over the Internet, entails the buildup of thousands of realtime connections between a transmitter and receiver programs, which are located e.g. in a personal computer of an Internet subscriber.
The currently most widely distributed Internet protocol http ensures an error-free transmission but does not possess any realtime properties. This protocol involves a network load at the transmitter which is proportional to the number of stations received. This means that the number of simultaneous listeners is typically limited to a few hundred, depending on the data rate of the coded signals, which is certainly not sufficient for “broadcasting applications”, i.e. applications in which radio or television signals are to be transmitted.
New protocols for transmitting data over the Internet, such as e.g. rtp (rtp=realtime transmission protocol), avoid this bottleneck in that, unlike the protocol http, they permit errors in the transmission.
This raises the problem for the receiver as to how it is to deal with erroneous transmission data. How will e.g. the abscence of parts of a multimedia or audio data stream be bridged?
DESCRIPTION OF PRIOR ART
Some methods for error management are already known in the prior art. In the method of mute switching, which is the simplest type of error concealment, the reproduction is switched off as long as data are missing or erroneous. The missing data are replaced by a null signal. Because of psychoacoustic effects, this sudden fall off and rise in the signal energy is felt to be very unpleasant. A mute switching method is described e.g. in Detlef Wiese: “Error Concealment Strategies for Digital Audio Broadcasting”, 92
th
AES Convention, Wien 1992, Preprint 3264, and in Detlef Wiese: “Anwendung von Fehlerverschleierungstechniken für digitalen Hörrundfunk (DAB)”, 9
th
ITG-Fachtagung “Hörrundfunk”, Mannheim 2/92.
Another known method is the method of data repetition. If a data stream ceases, a part of the data transmitted last is repeated in a loop. In block-oriented methods, which constitute the majority of known audio coding methods, the block or blocks transmitted last would be repeated for example. Such a method is known from the literature sources already cited.
This method leads to annoying secondary manifestations or artefacts, however: if only short parts of the audio signal are repeated, the repeated signal—whatever the original signal may be—sounds machinelike with a fundamental frequency at the repetition frequency. If longer parts are repeated, “echo effects” result, which are also felt to be annoying.
A further method, the method of data interpolation, is based on the probability that, when a data stream is transmitted with relatively high delay, valid audio data are already present again if an error gap of the audio data signal has to be concealed. An interpolation can be used to generate data in the gap, as described in Laurent Filliat, Mario Rossi, Joseph Maisano, “Error Correction by Interpolation in Digital Audio”, 92. AES Convention, Wien 1992, Preprint 3281.
A disadvantage of this method is e.g. that the delay is unacceptable in many cases, in particular in connection with telephony applications.
In block-oriented transform coders/decoders the signal energy per spectral line can be predicted or “predicated” from one block to the next. This known method (Jürgen Herre: “Fehlerverschleierung bei spektral codierten Audiosignalen”, Dissertation, Universität Erlangen-Nürnberg, Erlangen 1995) does indeed result in good error concealment, but a relatively high computational effort is required which makes a realtime decoding of a received multimedia or audio data signal impossible at the present time.
Also known in the technological field is the practice of increasing the redundancy in the data stream. It is possible to introduce into an audio data stream at the outset, in the coder or transmitter, a second data stream which transmits the same signal (possibly with a lower data rate and correspondingly lower quality). If this second data stream is displaced in time relative to the first, it is probable that at each moment valid data will be received from at least one of the two streams. The loss of one signal can thus be bridged by the other signal. In the case of an erroneous main signal this method provides a true reproduction of the information which, however, will normally be of worse quality.
As does the previous method, this method increases the delay between transmitter and receiver, so that the same disadvantages result as in the case of data interpolation. Furthermore, the data rate is increased, since a second, to some extent identical signal is transmitted, which cannot be acceptable especially at low data rates or low available transmission bandwidths.
For the transmission of a stereo signal there exists a further method for concealing errors in the transmission. In the method of left/right replacement the disturbed or failed channel can be replaced by the respective other channel as is described in Detlef Wiese: “Error Concealment Strategies for Digital Audio Broadcasting”, 92. AES Convention, Wien 1992, Preprint 3264, Detlef Wiese: “Anwendung von Fehlerverschleierungstechniken für digitalen Hörrundfunk (DAB)”, 9. ITG-Fachtagung “Hörrundfunk”, Mannheim 2/92, and Jürgen Herre: “Fehlerverschleierung bei spektral codierten Audiosignalen”, Dissertation, Universität Erlangen-Nürnberg, Erlangen 1995. In general, however, the data for both channels are transmitted simultaneously and are therefore disturbed at the same time. Otherwise what is involved here is a special case of the method for increasing the redundancy in the data stream. This method can, moreover, only be adopted when the left and the right channel can be decoded independently of one another. This is not possible, however, for the so-called “joint stereo” modes, which are defined by the Standard MPEG Layer 3, for instance.
SUMMARY OF THE INVENTION
It is the object of the present invention to provide a method for concealing errors in an audio data stream that attains an effective error concealment and that is as inaudible as possible.
This object is achieved by a method for concealing errors in an audio data stream, comprising the steps of detecting the occurrence of an error in an audio data stream, the audio data prior to the occurrence of the error being intact audio data; calculating a spectral energy of a subgroup of the intact audio data; forming a pattern for substitute data on the basis of the spectral energy calculated for the subgroup of the intact audio data; and creating substitute data for erroneous or missing audio data which correspond to the subgroup on the basis of the pattern.
A subgroup in the sense of this application always comprises several sequential frequency lines. The frequency domain is thus divided into sequential subgroups, each subgroup having sequential frequency lines. Preferably the subgroups can correspond to the frequency groups known from the field of psychoacoustics. It is also possible to choose a division for the subgroups which differs from the frequency groups, however, as will be described later.
As will become clear from the later detailed description, this method overcomes disadvantages occurring in the prior art since it neither requires an increased delay nor an increase in the data rate, i.e. a change in the source coding, nor exibits the unattractive time and frequency structure of a simply repeated signal section. In contrast e.g. to the method of spectral prediction, which has already been touched on, the method according to the present invention provides a hearing-adjusted noise substitution. The starting point is a temporal audio signal frequency representation which can be obtained in many ways which are known to persons skilled in t
Brandenburg Karlheinz
Gerhäuser Heinz
Herre Jürgen
Schildbach Wolfgang
Sieler Martin
Baker Stephen M.
Fraunhofer-Gesellschaft zur Forderung der ange-wandten Forschung
LandOfFree
Method for masking defects in a stream of audio data does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for masking defects in a stream of audio data, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for masking defects in a stream of audio data will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2847670