INTERNATIONAL ORGANIZATION FOR STANDARDIZATION
`
`ORGANISATION INTERNATIONALE DE NORMALISATION
`
`ISO/I EC JTCI/SC2/WG8
`
`CODED REPRESENTATION OF PICTURE AND AUDIO INFORMATION
`
`ISO-IEC/JTC1/SC2/WG8
`MPEG 89/160
`July1989
`
`Source
`
`Title
`
`Status
`
`Hans-Georg Musmann
`Chairman of MPEG/Audio
`Stockholm Meeting Report
`Separate Meeting of the MPEG/Audio Group
`Draft
`
`Separate Meeting of the MPEG/Audio Group
`
`the MPEG/Audio group discussed audio related
`from July
`During the three days
`to July
`problems in separate sessions The list of participants of this session is given in Annex
`
`On the first day paragraph
`performances to be tested of the Proposal Package Description
`Doc MPEG 89/128 was analysed in order to identify interdependencies
`between audio video
`VLSI and system aspects
`
`list of items concerning audio related system aspects was elaborated and forwarded
`input to the SYSTEM Group The list
`is given in Annex II
`The VLSI questionnaire Doc MPEG 89/120 was discussed in order to find out whether
`there are aspects related to audio coding which are not reflected well
`The following recommendation was stated and forwarded to the VLSI Group The VLSI
`questionnaire should consider aspects of floating point arithmetic
`
`as
`
`On the second day the participants had
`discussion about the 14 submitted proposals for audio
`coding algorithms The top level descriptions of
`are contained in document
`these algorithms
`MPEG 89/148 The results of
`in the new draft version of the
`the discussion
`are given in detail
`Proposal Package Description Doc MPEG 89/128 paragraph 9.2.1 Preselection rules sound
`
`The next point of the discussion was
`revision of the test procedure The results of this discussion
`were also used as input for the new draft version of the Proposal Package Description Doc MPEG
`89/128 and can be found in paragraph
`Performances to be tested paragraph 8.2 Definition
`test procedure sound and paragraph 9.22 Performances
`of
`to be tested
`for preselection
`sound
`
`to the evaluation
`
`of
`
`With respect
`the test
`the chairman proposed
`results
`procedure which is described
`in Annex ifi The goal of
`evaluation
`this procedure is to find out
`whether one algorithm is suitable for broadcast needs as well as for all other needs or not This
`proposal has to be discussed during the October meeting
`
`hierarchical
`
`Based on document MPEG 89/117 List of proposed test sequences
`for subjective testing
`twenty test sequences were selected for subjective testing The complete list of these twenty test
`separate document MPEG 89/166
`is given in
`In the case that algorithms will be
`sequences
`
`1
`
`HP 1025
`
`

`

`tested which exploit
`the stereo channel
`test procedure are needed
`for
`proposals
`
`interdependency
`
`some additional
`
`test sequences
`
`and
`
`document provided by Dr Chiariglione
`and directed to the Mvisoiy Group Multimedia
`Coding Activities in Italy raised
`discussion
`about
`the problemof dissemination of detailed
`descriptions of algorithms and software The audio group agreed on the following recommenda
`tions
`
`Detailed descriptions of algorithms and software should not be disseminated
`should be directed to the originator
`in written form
`Requests
`
`Furthermore there were some concerns about
`the situation of non- disclosure agreements
`within the ISO In the absence of exact knowledge
`it was recommended that members who want
`to impose non- disclosure agreements should distribute the forms for agreements early before the
`October meeting In addition the following two questions were directed to the convenor of WG8
`by the chairman of the MPEG/Audio group
`for non-disclosure agreements within the ISO
`Are there regulations
`Is there any regulation
`for rejecting parties who insist on
`non-disclosure agreement
`
`The answers wifi be discussed during the October meeting
`
`Since
`
`several members of the audio group are involved
`in other conferences in October
`proposed to start with the separate meeting of the MPEG/Audio group on October
`possible to finish on October 20 or 21
`
`it was
`18 and if
`
`The members of the audio group are asked to prepare proposals
`for an appropriate
`procedure and statistical evaluation of the A-B-X test by the October meeting Furthermore they
`are asked to submit proposals
`for an appropriate
`quality score related to the A-B-X test until
`September30 see also new version of document MPEG 89/128 Proposal Package Description
`Finally proposals for locations of the subjective tests or volunteers
`for hosting
`requested until the October meeting
`
`these tests are
`
`test
`
`Mnex
`
`2
`
`

`

`Stockholm meeting attendance fist MPEG Audio
`CCETF
`Philips CE
`Dolby Laboratories Inc
`Universitaet Hannover
`
`NL
`USA
`
`Deheiy
`Eisema
`
`Fielder
`
`Fuchs
`
`USA
`GB
`
`Gerhaeuser
`
`Jayant
`Kim
`Kohnert
`Mahieux
`Musmann
`Noll
`
`Spille
`
`Stoll
`Taka
`
`Theile
`Uwabu
`
`Fraunhofer Gesellschaft
`ATT Bell Laboratories
`
`British Telecommunications
`plc
`German Aerospace Research Establishm
`CNET France Telecom
`Universitaet Hannover
`
`Institut
`
`Universitaet Berlin
`Deutsche Thomson-Brandt GmbH
`flier Rundfunktechnik GmbH
`Nippon Telegraph and Telephone Corp
`flier Rundfunktechnik GmbH
`Institut
`Graphics Communication Technologies
`
`Annex II
`
`3
`
`

`

`List of items to be considered
`by the SYSTEM Group
`Numbers are with respect
`to paragraph Tot document MPEG 891728
`
`7.1 Motion video decoding playback
`The maximum allowed decoding delay for audio is 0.5 sec
`Further
`information concerning the delay between audio and video is required
`
`7.2 Wideband audio decoding playback
`It has been agreed to use the following definitions of terminology
`audio block refers to signal processing unit
`audio frame refers to access unit
`An audio frame may consist of several audio blocks
`
`7.4 Random access
`Different access units for video and audio are assumed
`This assumption arises the question whether there is synchronization necessary between
`audio and video access units
`
`7.10 Data capacity for ancillary information
`The capability
`for audio related ancillary data has to be provided by the audio coding
`The maximum expected bit rate for audio related ancillary data is bit per every 64 bits
`of the transmission bit stream
`In the case of audio all subjective tests will be performed with zero ancillary data
`
`transcoding
`
`of CCrfl px64 kbitls video
`
`7.11 Direct
`and
`7.13 Transcoding to AES/EBU consumer.standards and CC1TF video conferencing standards
`Transcoding of the video signal has to consider transcoding problems of the audio signal
`In the case of audio the problems of transcoding will not be part of the proposal package
`
`7.15 Synchronization of audio and video
`and
`7.17 Ability to do insert/cut
`see 7.4
`
`editing audio
`
`7.22 Robustness to bit errors in storage/transmission
`The SYSTEM Group is asked to provide error patterns in the decoded bit stream after
`error correction and error randomization de-interleaving
`Uncorrectable
`but detectable block errors should be indicated by flags
`
`7.23 Encoder cost and/or complexity
`In view of recording aspects the SYSTEM Group is asked to consider encoder cost and/or
`complexity
`
`7.26 Short decoding delay
`see 7.1
`
`7.27 Short encoding delay
`Encoding delay for audio has to be less than 03 sec
`
`4
`
`

`

`Annex Ill
`
`Proposal from the chairman
`evaluation procedure
`Chierarchicaf
`
`The goal of this procedure is to find out whether one algorithm is suitable for broadcast needs
`as well as for all other needs or not
`
`With respect
`to the broadcast needs the main performance is the audio quality which has the
`highest weight wi
`
`Assume that
`
`there will be
`
`performances
`
`to be tested and the following correspondence
`
`is
`
`defined
`
`Index
`
`corresponding performance/implication
`
`implication with highest weight
`implication with second highest weight
`
`implication with weight wi
`
`implication with lowest weight
`
`sequential evaluation
`
`of the tests and ranking of the algorithms will be applied in the following
`
`way
`
`SSixwi
`
`SS1xwiS2xw
`
`SS1xw1-i-S2xw2...S1xw
`
`SSlxwlS2xw2...Sjxw1...SKxwK
`
`where
`
`score of an algorithm
`score of an algorithm for implication
`weight of implication
`
`Si
`
`wi
`
`At each step all algorithms are ranked according to the score
`
`As long as one of the algorithms which satisfy the audio quality needs of the broadcasters
`is on
`top of the ranking solution can be found which is suitable for all needs Otherwise
`two solutions
`have to be considered
`
`This proposal has to be discussed during the October meeting
`
`5
`
`