`
`US005659660A
`[HJ Patent Number:
`[451 Date of Patent:
`
`5,659,660
`Aug. 19, 1997
`
`United States Patent c191
`Plenge et al.
`
`[54] METHOD OF TRANSMITTING AND/OR
`STORING DIGITIZED, DATA-REDUCED
`AUDIO SIGNALS
`
`[75]
`
`Inventors: George Plenge, Thanning; Dedef
`Wiese, Hallbcrgmoos; Martin Uok,
`Munich, all of Germany
`
`[73] Assignee: Institut fuer Rundfunktechnik
`GmbH, Munich, Germany
`
`(21] Appl. No.:
`
`313,255
`
`(22] PCI' Filed:
`
`Mar. 22, 1993
`
`(86] PCI'No.:
`
`PCT/EP93100690
`Jan. 3, 1995
`§ 371 Date:
`§ 102(e) Date: Jan. 3, 1995
`[87] PCT Pub. No.: W093/21694
`
`PCT Pub. Date: Oct. :28, 1993
`
`(30]
`
`Foreign Application Priority Data
`
`[DEJ Germany .......................... 42 11 945.6
`
`Apr. 9, 1992
`Int. Cl.6
`.. _ ..................... -·-··········- .......... - GlOL 9/00
`[51]
`[52] U.S. Cl .......................... 395/2.29; 395/2.1; 39512.14;
`395/2.22; 395/2.21
`(58] Field of Search ___ ........................... 39512.2, 2.21,
`395/2.22, 2.14, 2.29; 381/34
`
`(56]
`
`References Cited
`
`U.S. PAlENT DOCUMENTS
`
`3,023;2.77
`3,621,3!n
`4,038,495
`4,370.643
`4,546,342
`4,821,200
`4 ,882,754
`4,942,607
`5,115,469
`
`211962 Mathews ········-···-········· .. ·······• 395/2.2
`1111971 Kanagawa et al. ··-·-····-···· ... 39512.21
`711977 White ····-···-··-······················ 395/2.29
`111983 Kitamura ···-··-···-·-.............. 395/2.21
`10/1985 Weaver et al ---·-·-··········-···· 341/51
`4/1989 Klank et al. ··--·-·-····-·-· .. ·• 370/345
`11/1989 Weaver et al . ........................... 395/2.1
`711990 Schroder et al. ...................... 395/2.12
`5/1992 Taniguchi et al . ..................... _ 381/34
`
`5,179,623
`
`1/1993 Dickopp et al. ····---········· 39512.09
`
`FOREIGN PATENT DOCUMENTS
`
`.......... H04B 1/66
`9/1986 European Pat. Off .
`0193143 A2
`0290581 Bl 11/1988 European Pat. Off ........... H04B 1166
`7/1990 European Pat. Off ..
`0376553A2
`3629434 Al
`3/1988 Germany ........................ H03M l.W
`OTHER PUBLICATIONS
`Lee, Francis F. ("Time Compression an!l Expansion of
`Speech by the sampling Method", 42nd Convention of the
`Audio engineering society, pp. 738-742, May, 1972). May
`4, 1972.
`
`Primary Examiner-Allen R MacDonald
`Assistant &aminer-Vijay B. Chawan
`Atto171e)I Agent, or Finn-Spencer & Frank
`
`[57)
`
`ABSTRACT
`
`A method of encoding an audio signal for transmission
`and/or for storage includes changing a digitized audio signal
`based on a selected signal changing criteria. The changed
`digitized audio signal is encoded, based at least on an
`available data rate for transmitting and/or storing the audio
`signal, for reducing an amount of data associated with the
`changed, digitized audio signal. The reduced-data audio
`signal is then evaluated for determining whether a code
`overload condition exists. H a code overload condition
`exists, another signal changing criteria is selected. A time
`behavior and/or a spectral distrlbution of the digitized audio
`signal is recursively changed based on the another signal
`changing criteria before the amount of data associated with
`the digitized audio signal is reduced so that deterioration of
`reduced-data audio signal is not perceptible when the coded
`signal is decoded. The recursively changed digitized audio
`signal is encoded for reducing an amount of data associated
`with the recursively changed digitized audio signal based at
`least on the available data rate. Lastly, the recursively
`changed encoded reduced-data digitized audio signal is
`transmitted and/or stored.
`
`1t Claims, 1 Drawing Sheet
`
`14
`
`·22
`Signal
`Change
`Stage
`
`33
`34
`Doto
`Reduction ...------
`
`10
`11 A-D
`Converter
`
`12
`
`Dato
`Expansion
`
`50
`
`40
`
`Signal analysis
`
`51
`
`IPR2016-01710
`UNIFIED EX1016
`
`
`
`U.S. Patent
`
`Aug. 19, 1997
`
`5,659,660
`
`14
`
`22
`Signal
`Change
`Stage
`
`33
`,
`34
`23 Dato
`Reduction ,__ _ __._~
`
`10
`1 1 A- 0
`Converter
`
`12
`
`20
`Signal
`Change
`St age
`
`41
`
`f 1
`
`43
`
`32
`
`40
`
`30
`Data
`Reduction
`(encoder)
`
`1
`
`44
`
`----·
`
`42
`
`Doto
`Expansion
`
`50
`
`Signal analysis
`
`51
`
`Figu re 1
`
`
`
`5,659,660
`
`1
`METHOD OF TRANSMITTING AND/OR
`STORING DIGITIZED, DATA-REDUCED
`AUDIO SIGNALS
`
`BACKGROUND OF THE INVENTION
`
`2
`DETAILED DESCRIPTION OF THE
`PREFERRED EMBODTh1ENTS
`In the encoder shown in FIG. 1, the arriving analog audio
`signal 11 is fed to an analog/digital conversion stage 10. The
`5 analog signal 11 has various signal features, such as a
`spectral distribution and a time behavior (or behavior during
`the course of time). The digitized audio signal 12 is fed to
`1. Field of the Invention
`a stage 40, to be explained later, for signal analysis, to a
`The invention relates to a method of transmitting and/or
`signal change stage 20, as well as to delay elements 15 and
`storing digitized, data-reduced audio signals. Such a method
`is known, for instance, from the European patent specifica- 10 70. In the signal change stage 20, the digitized audio signal
`is processed as a function of a change recommendation 41
`tion 290 581.
`2. Description of the Related Art
`generated by the signal analysis stage 40. The change
`In a bit rate-reduced encoding of audio signals that are
`recommendation 41 may also be to not change anything. The
`processed and possibly changed audio signal 21 at the output
`already available in a digitized form, e.g., having a 48 kHz 15 of stage 20 is reduced in a subsequent data reduction stage
`scanning frequency/16 bit resolution, utilization of psychoa-
`30 with regard to the data rate available for transmission or
`coustic phenomena in the perception of audio signals is
`known from EP 290 581 in a way that the original bit rate
`storage. This data reduction is carried out as a function of an
`is considerably reduced. Such methods carrying out
`encoding recommendation 42 which is made available by
`the signal analysis stage 40. The encoded audio signal 31 at
`transcoding from a higher to a lower bit rate are commonly
`known as source coding. They particularly utilize masking 20 the output of stage 30 is fed to a data expansion stage 50 for
`use in a recursive control process, which will be described
`effects of the ear in the frequency and time domain. It is the
`in greater detail below, in which the data-reduced audio
`object of these methods to not change, if possible, the
`signal is expanded so that, at the output of this stage 50, a
`audible quality which the audio signal has before transcod-
`digitized audio signal 51 is available, in a linearly coded 16
`ing ( so;:ce) coding) by reducing the bit rate (optimum 25 bit pulse code modulation. This re-expanded digital audio
`transco ng ·
`signal 51 is fed to the signal analysis stage 40. In addition,
`In many applications, there is a limited data capacity
`there are applied to the input of stage 40 the digitized audio
`~vailable for. trans~ss!on or ~torage. If this bit rate, w~ch
`signal 12 to be encoded as well as an information signal 32,
`is ma~e available, .1s .msufficient: a further da~ reduction
`supplied by the data reduction stage 30, concerning the data
`accor~ng to the_ pnncipl~s of optimum tran~co~g leads to 30 rate required for the encoding of the audio signal 12, with the
`some~es cons1derabl~ mt~erence: especi~y m the form
`data rate providing information on the existing code viola-
`of considerable soun~ distortions. Thi~ s~e mterference, of
`tions during the data reduction of the audio signal 12.
`course, also occ~s if the source coding is not two-staged,
`Furthermore, stage 40 is fed a differential signal 13 which
`but takes place m one step.
`results from the generation of the difference between the
`Furthermore, in a method of transmitting and/or storing 35 re-expanded audio signal 51 and the audio signal 12 to be
`digitized, data-reduced audio signals it is known (DE 36 29
`encoded. In order for the differential generation to relate to
`434 Al) to let further encoding steps follow if, after
`temporally synchronous audio signals, the audio signal 12 to
`encoding, a predetermined bit rate is exceeded (code
`be encoded is delayed by the amount At1 in the delay stage
`70 and is also fed to stage 40. Based on the information
`overload), until the actual bit rate no longer exceeds the
`designated bit rate. That is, if code overload exists after a 40 signal 32, the signal analysis stage 40 determines whether a
`first encoding sep, further encoding steps can be employed.
`code overload is present. A code overload is defined as a
`However, this approach does not always avoid a code
`state in which the available capacity of a transmission or
`overload with certainty.
`storage medium is not sufficient for the optimum transcod-
`ing of the audio signal. If such a code overload is present, the
`45 recursive control process is employed, with the input signals
`12, 13 and 51 applied to the signal analysis stage 40 being
`examined with regard to possible signal changes as well as
`to encoding strategies. Possible signal change measures, or
`measures for changing the signal changes or criterion, that
`50 are features of the audio signal (unless the signal change
`measure indicates no change), communicated to the signal
`change stage 20 in the form of a change recommendation 41,
`limitation or a time behavior change such as a level change
`of the audio signal to be encoded, which lead to a reduction
`55 of the data rate required in the data reduction stage 30 and,
`at the same time, to a minimum interference effect of the
`re-expanded audio signal 51. Apart from the bandwidth
`limitation and level change mentioned, further typical signal
`change measures include other spectral distribution changes
`60 such as frequency response changes (pre-emphasis), and
`other time behavior changes such as steeper or lesser signal
`amplitude changes, or signal repetitions in place of the
`actual signal, particularly for periodic or quasi-periodic
`signal characteristics of the audio signal. Simplification of
`65 the spatial imaging or distribution of multichannel audio
`signals, such as providing mono instead of stereo for the
`entire audio frequency range or for a part of the frequency
`
`In contrast, it is the object of the present invention to
`minimize the interference or distortion effect during a fur(cid:173)
`ther data reduction of an already data-reduced audio signal
`or during a first data reduction of an audio signal, while
`accepting code violations.
`This object is provided by the present invention so that in
`the event of a code overload, before the audio signal is
`encoded, the time behavior and/or the spectral distribution
`of the audio signal is changed according to a selected
`adaptation pattern without a considerable signal deteriora(cid:173)
`tion being subjectively perceivable after decoding the
`encoded audio signal.
`Advantageous modifications of the solutions according to
`the invention are described as follows.
`
`SUMMARY OF THE INVENTION
`
`BRIEF DESCRIPTION OF THE DRAWING
`
`By way of an embodiment, the invention is explained in
`greater detail in the only drawing. It shows:
`FIG. 1 is a block diagram of an encoder for the imple(cid:173)
`mentation of the method according to the invention.
`
`
`
`5,659,660
`
`3
`range, may also be used as a signal change measure. A signal
`change measure selected by the signal analysis stage 40 is
`communicated to the signal change stage 20 in the form of
`a change recommendation 41. The signal analysis stage 40
`also generates an encoding recommendation 42 which is 5
`communicated to the data reduction stage. The encoding
`recommendation 42 is selected from a table of distortion or
`interference effect patterns in the signal analysis stage 40 by
`using the change recommendation 41 generated there. The
`encoding recommendation 42 is used in the data reduction 10
`stage 30 for a transcoding of the audio signal to be encoded
`at a reduced bit rate so that the interference or distortion
`produced will be as small as possible. The recursive control
`process required for the signal analysis in stage 40 can also
`be carried out several times, if necessary, until a satisfactory 15
`result is achieved by the change and encoding recommen(cid:173)
`dations 41, 42. As will be apparent from the foregoing, if a
`code overload condition is detected then at least one signal
`feature of the audio signal is recursively changed. For the
`final signal change and data reduction, the final change and 20
`encoding recommendations 43 or 44 are fed to a further
`signal change stage 22 or a further data reduction stage 33,
`which function in the same manner as the stages 20 and 30.
`Since a certain amount of time is necessary for the recursive
`control process, the audio signal 12 must be time-delayed by 25
`the amount ~~ before the final suboptimum transcoding by
`means of stage 15. The data-reduced audio signal 34 result(cid:173)
`ing from the data reduction stage 32 can be transmitted, e.g.,
`to a channel encoder, which is not shown, for storage on a
`sound carrier or for the processing of a broadcast signal. 30
`During the encoding of the audio signal, code bits are
`assigned to the data-reduced encoded signal based on the
`recommended changes made to the digitized audio signal.
`When very a low bit rate of the data-reduced audio signal
`is available, the pwposeful change of the digital signal prior 35
`to encoding is effected so that the speech intelligibility of the
`audio signal is maintained or is improved, if possible.
`During a stereophonic encoding, or when two or more audio
`signals are encoded together or as a function of one another,
`the purposeful change of the various audio signals occurs
`prior to the data reduction/encoding of the signals, with the
`change being based on mutual level relationships of the
`audio signals. The selection of the particular encoding
`scheme and/or the particular pwposeful change of the audio 45
`signal prior to encoding are/is respectively determined anew
`after short periodic time intervals.
`The method according to the invention can also function
`without recursive control process if an internal examination
`of the audio signal 51 which has been data-reduced and 50
`re-expanded by the change and encoding recommendations
`41 or 42 is dispensed ~ith. The advantage related herewith
`of reduced physical intricacy (complexity), however, is
`gained at the expense of the achievable optimum of the
`suboptimum transcoding. If the recursive control process is
`dispensed with, the input signals 13 and 51 at the signal
`analysis stage 40, as well as the time-delayed audio signal
`12, are deleted. The already encoded signal 31 represents the
`final encoded audio signal so that also the stages 15, 22 and
`33 are not required. A further simplification of the method
`according to the invention is achieved by, in addition,
`dispensing with the continuously conducted signal analysis,
`by, e.g., setting the signal change stage 20 to a constant
`change and the data reduction stage 30 to a constant encod(cid:173)
`ing recommendation.
`
`4
`It is understood that the signal processing that may have
`to be made in the signal change stage 20 can also be carried
`out prior to the analog/digital conversion of the analog audio
`signal 11. For this purpose, the stage 20 may be re-located
`so that it is connected in series before the analog/digital
`converter.
`We claim:
`1. A method of encoding an audio signal for transmission
`and/or for storage, the method comprising the steps of:
`changing a digitized audio signal based on a selected
`signal changing criteria;
`encoding the changed digitized audio signal for reducing
`an amount of data associated with the changed digitized
`audio signal based at least on an available data rate for
`transmitting and/or storing the audio signal;
`evaluating the reduced-data audio signal for determining
`whether a code overload condition exists;
`selecting another signal changing criteria if a code over(cid:173)
`load condition exists;
`recursively changing a time behavior and/or a spectral
`distribution of the digitized audio signal based on the
`another signal changing criteria before the amount of
`data associated with the digitized audio signal is
`reduced so that deterioration of reduced-data audio
`signal is not perceptible when the recursively changed
`digitized audio signal is decoded after transmission
`and/or storage;
`encoding the recursively changed digitized audio signal
`for reducing an amount of data associated with the
`recursively changed, digitized audio signal based at
`least on the available data rate; and
`transmitting and/or storing the recursively changed
`encoded reduced-data digitized audio signal.
`2. The method according to claim 1, wherein the step of
`selecting another signal changing criteria comprises the
`steps of:
`expanding the reduced-data digitized audio signal;
`generating a difference signal between the digitized audio
`signal and the expanded digitized audio signal; and
`evaluating the digitized audio signal, the difference signal
`and the expanded digitized audio signal for selecting
`the another signal changing criteria.
`3. The method according to claim 2, further comprising
`the steps of:
`evaluating the digitized audio signal, the difference signal
`and the expanded digitized audio signal for selecting an
`encoding strategy for encoding the recursively changed
`digitized audio signal; and
`selecting the encoding strategy based on a table of inter(cid:173)
`ference effect patterns.
`4. The method according to claim 3, further comprising
`the step of delaying the digitized audio signal for a prede-
`55 termined period of time before recursively changing the
`digitized audio signal.
`5. The method according to claim 2, wherein the another
`signal changing criteria includes at least one of a frequency
`response change, a temporal change, and a signal repetition
`60 in place of an actual signal.
`6. The method according to claim 2, wherein the another
`signal changing criteria includes a simplification of a spatial
`imaging of multichannel audio signals when the audio signal
`65 comprises a plurality of audio signals.
`7. The method according to claim 1, wherein when the
`audio signal is a stereophonic signal or two or more audio
`
`40
`
`
`
`5,659,660
`
`5
`signals, at least one of the steps of changing the digitized
`audio signal and recursively changing the digitized audio
`signal is based on mutual level relationships of the audio
`signals.
`8. The method according to claim 1, wherein when the 5
`available bit rate is a low bit rate, at least one of the steps of
`changing the digitized audio signal and recursively changing
`the digitized audio signal causes a speech intelligibility of
`the audio signal to at least be maintained or to be improved.
`
`6
`9. The method according to claim 1, wherein at least one
`of the steps of encoding comprises the step of assigning code
`bits based on a previous change of the digitized audio signal.
`10. A method according to claim 1, further comprising the
`step of repeating the steps of the method after a predeter(cid:173)
`mined period of time.
`
`* * * * *