CONFERENCE REPORT

CONFERENCE REPORT

788 J. Audio Eng. Soc., Vol. 62, No. 11, 2014 November

55th International Conference

Spatial Audio

27–29 August 2014Helsinki, Finland

Chair: Lauri Savioja

CONFERENCE REPORT

J. Audio Eng. Soc., Vol. 62, No. 11, 2014 November 789

At the end of August 2014, the Audio Engineering Society was welcomedback to Helsinki for the 55th Conference. It was the sixth AES conferenceto be held in Finland, and the topic was also an AES staple: spatial audio.

There have been seven conferences directly concerned with spatial audio and nodoubt many more related sessions, workshops, and tutorials at other events. Theconference brought together a diverse range of audio scientists from academiaand industry (with a 50% split between the two), with interests ranging from thetechnical reproduction of sound fields, perceptual evaluation, product/servicedesign, musical performance, and more. The wide range of participants (130attendees from 25 countries) helped to foster a positive atmosphere of discus-sion, ideation, and collaboration.

The conference took place in the Helsinki Music Centre, which opened in 2011and is shared by the Sibelius Academy, the Finnish Radio Symphony Orchestra,and the Helsinki Philharmonic Orchestra. The Centre contains a range of per-formance spaces (the main auditorium was designed by Yasuhisa Toyota andseats 1,700), rehearsal rooms, and recording studios.

Sponsors

CONFERENCE REPORT

The conference organizing committee should be commended forhosting such a successful and engaging event, curating an enter-taining program of technical and social events. The conference waschaired by Lauri Savioja, Ville Pulkki compiled the technicalprogram, Tapio Lokki acted as papers chair, Teemu Koski was thesecretary, Andrew Goldberg organized sponsorship, Julia Turku wastreasurer, Kalev Tiits was in charge of the venue, and FlorianCamerer helped to organize the panels.

SEMINARS AND DEMONSTRATIONS AT AALTO UNIVERSITYProceedings started a day before the conference as 85 participantswere treated to an extra afternoon of tutorials and demonstrationshosted by the acoustics group at Aalto University. Ville Pulkkiopened the event with a talk about psychoacoustics and technolo-gies of spatial sound. The presentation was intended to offer someconclusions and output at the culmination of a five-year project thathad recently finished. The talk focused on the relationship betweenengineering and psychoacoustics, bridging the gap between spatialsound technology and psychophysics. Pulkki’s group was taskedwith creating a flexible and loudspeaker-agnostic generic audio for-mat: this was the motivation behind the “DirAC” (directional audio

coding) system, which analyzes and separates directional informa-tion and diffuse sound across frequency bands and uses this infor-mation in the resynthesis of spatial sound. Pulkki talked about vari-ous problems that had been encountered and the resulting solutionsthat were implemented in order to improve the performance of thesystem. The engineering work on DirAC was complemented by per-ceptual investigation of aspects such as the perception of spatially-wide sources and the relationship between audio and video spatialfield width.

Following this talk, participants were divided into groups andshown around eight demonstrations in the impressive acousticsfacilities at Aalto University. Demos included an immersive audio-visual environment for teleconferencing with 226-degree-horizon-tal-field-of-view video and 3D audio using DirAC, 12-channelauralizations of concert halls and studio control rooms, and a head-tracked Oculus Rift AV experience, among other examples of tech-nical aspects of the DirAC system.

The afternoon was rounded off with a talk by Tapio Lokki whodetailed his work on the evaluation of concert hall acoustics,considering the mapping between physical factors, sensory evalua-tion, and preference. The talk provided an interesting overview ofsensory evaluation methods, which are often borrowed from foodand wine science and can include individual assessors developingtheir own descriptive vocabulary with the use of advanced statisti-cal methodologies to assess the relationship between physicalparameters, perceptual ratings, and the assessors. Lokki also gavean interesting description of his tour of European concert halls inorder to capture room impulse responses: “we stole the acoustics,packed it, and put it in a bus.”

KEYNOTELECTURESEach of the three days ofthe conference startedwith an invited presenta-tion from a distinguishedresearcher in the field,carefully selected by theorganizing committee tocover the engineering,perceptual, and musicalaspects of spatial soundtechnologies.

On day one, SaschaSpors gave a talk entitled “The Adventureof Spatial SoundReproduction,” whichwas the perfect way tokick off the conference.The talk included an outline of the history of spatial audio repro-duction, starting with the invention of the telephone in 1861 allthe way through to modern multichannel systems and panningalgorithms. Sascha touched upon a number of points that would berevisited in presentations throughout the conference: the fact thatit is often necessary to create the desired impression rather than toexactly duplicate a real listening experience; the fact that it isimportant not to overlook timbral factors in a spatial audio system;and the relationships between signal processing, perception, andacoustics, which when integrated give a complete view of spatialsound reproduction. During the questions after the talk partici-pants were introduced to the “Catchbox,” a throwable microphone

790 J. Audio Eng. Soc., Vol. 62, No. 11, 2014 November

Members of the AES 55th Conference organizing committee: from left,Julia Turku, Ville Pulkki, Tapio Lokki, Lauri Savioja, Andrew Goldberg,and Teemu Koski

Sascha Spors identifies some key areasof spatial audio during his keynotelecture.

Sakari Tervo presents a demonstration of auralizations of concert hallsand studio control rooms in the largest of three anechoic chambers atAalto University.

CONFERENCE REPORT

that made it rela-tively easy for themicrophone to bemoved around toauditorium, as wellas providing asource of entertain-ment as the softbox was flungacross the room.

Jonty Harrison’skeynote talk on thesecond day was agood example of

the diversity of participants and the outlook of the AES. As an“acousmatic” composer (i.e., music that is made in a studio,performed over loudspeakers, exists only as a recording, and has anunlimited sound palette), Harrison described his modus operandias being flexible, listening based, and not based around prior plan-ning, and suggested that audio engineering can facilitate thisapproach (for example, such composition is only possible due tothe development of sound recording and storage). Harrison intro-duced the loudspeaker layouts used in the Birmingham BEAST(Birmingham ElectroAcoustic Sound Studio) project, which hasover one hundred loudspeakers at its disposal for composition andperformance. It was interesting to note that the motivation forloudspeaker positioning was empirical and listening-based, andthat in this particular use of spatial sound technology, it wasimportant to create the sensation of space and movement ratherthan extreme accuracy of spatial audio field reproduction.

The final invited presentation was given by Piotr Majdak, whospoke about “sound localization beyond the horizontal plane.”Majdak spoke about localization mechanisms in humans, startingfrom the early work by Lord Rayleigh to more modern theories andmethods. Binaural methods and head-related transfer functions(HRTFs) were the subject of much discussion during the confer-ence, particularly the need for individualized HRTF measurements;Majdak discussed the search for a metric for evaluation of HRTFsand detailed work that had considered the “spectral difference”between a pair of measurements.

The three invitedpresentations served as agood background for thedetailed investigations ofthe substantial paper andposter program, whichfeatured 24 presentationsand 16 posters.

SPATIAL SOUNDTECHNIQUESDay one of the confer-ence featured two papersessions on spatial soundtechniques,”chaired bySakari Tervo (Aalto University).

P h i l i p p e - A u b e r tGauthier from the University of Sherbrooke spoke about the benefitsof accurate simulation of spatial sound fields when investigatingsafety in industrial environments and the need for nondisruptivemethods for capturing such environments. He introduced a combi-nation of spot microphones and microphone arrays that were used toseparate foreground and background sounds. Simulation resultswere promising and the work was due to be continued with realrecordings reproduced over wave field synthesis (WFS). The follow-ing paper, presented by Symeon Delikaris-Manias of Aalto University,also considered a method for separating acoustic sources, introduc-ing modifications to the cross pattern coherence algorithm(CroPaC), which had been demonstrated at Aalto the previous day.The next two papers concentrated on methods for reproducing soundfields. Philip Coleman (University of Surrey) introduced planaritypanning as a way of producing a spatial sound field. This methodfocuses sound energy on a target zone while also constraining theangle of arrival and was shown to accurately position five virtualloudspeakers. Naomici Yanagidate (Universities of Southampton andChuo) compared methods for generating personal sound zones in acar cabin, proposing a method based on acoustic contrast controlthat also constrained the variance in sound pressure throughout thetarget zone and presenting simulation results.

J. Audio Eng. Soc., Vol. 62, No. 11, 2014 November 791

Philip Jackson demonstrates his catchingskills to ask a question using theCatchbox microphone.

The conference attracted 130 participants from 25 countries.

MorrowSound ran a 3D audio-visual demoover Occulus Rift during the conference.

The session continued in the afternoon with two papers aboutHRTFs. Bosun Xie (South China University of Technology)presented a method for simplifying the personalization of HRTFs,analyzing similarity using a Wavelet decomposition, and DineshManocha (University of North Carolina) presented a fast methodfor computing personalized HRTFs taking only 20 minutes on adesktop computer.

The spatial sound techniques session continued on day two,with the first paper presented by Andreas Franck (Fraunhofer),who spoke about an efficient frequency-domain crossfadingmethod for interpolation of HRTFs, which is required for binauralsynthesis of dynamic acoustic scenes. The following papersfocused on loudspeaker-based reproduction methods, with DavideScaini (Barcelona Media) and Tobias Weller (MacquarieUniversity) talking about ambisonics followed by another presen-tation from the University of Sherbrooke by Anthony Bolduc whoused WFS for sound field reproduction of vibroacoustic models.The last paper of the spatial sound techniques sessions waspresented by Mikko-Ville Laitinen (Aalto University), whoproposed a modified amplitude-panning law with a frequency-dependent gain normalization factor for ensuring the constantperceived loudness of a panned source, generating many ques-tions from the interested audience.

SPATIAL SOUND ENGINEERINGThe spatial sound engineering session contained four papers with arange of technical and operational outlooks. Panagiotis Char-alampous (Cyprus University of Technology) compared varioustree-traversal algorithms, which are an integral part of geometricalacoustic methods of sound propagation simulation. Results sug-gested that best-first methods are good for real-time applications,while depth-first methods are good for offline. Akio Ando (Univer-sity of Toyama) gave an in-depth account of Makita’s 1962 theory ofsound localization and built upon this to derive a model of soundlocalization in multiloudspeaker reproduction. Jose J. Lopez (Uni-versity of Valencia) presented a more application-based paper,introducing a system for teleconferencing over a mobile phone thatgave the user various spatialization options in order to improveclarity and immersion in a conference call. Finally, Francis Stevens(York University) presented a method for capturing spatial impulseresponses, comparing between swept-sine and starter-pistol excita-tion methods and suggesting the benefits of each.

SPATIAL SOUNDPSYCHOPHYSICS The final two paper sessions weretitled Spatial Sound Psychophysics.Presentations delved into a range ofissues regarding perception of spa-tial sound. The first paper, pre-sented by Alexander Adami (Inter-national Audio Laboratories,Erlangen), considered down-mix-ing. A MUSHRA test was used tovalidate the performance of a newmethod in which coherent signalparts are identified and suppressedin one channel. The method wasfound to be preferred when resultswere aggregated, although therewere some pronounced programdependencies. Olli Santala (AaltoUniversity) presented some detailed

results of listening tests performed with pulse sequences distributedover an array of loudspeakers, suggesting that spatio-temporal hear-ing resolution is surprisingly low. Hagen Wierstorf (TU Berlin) inves-tigated coloration in various WFS arrangements, again stressing theimportance of timbral quality in a spatial audio system. Michael Sco-effler (International Audio Laboratories, Erlangen) looked into theinfluence of replay systems and preference for the audio material onthe overall listening experience, and grouped listeners into two cate-gories: “song likers” and “multichannel likers.” The last paper beforelunch saw a return to HRTFs with Kyla McMullen (Clemson Univer-sity) presenting results about the consistency of selection of HRTFsover time; listeners were found to be fairly consistent in their selec-tions, leading to a need to study similarities between selected HRTFs.

Frank Melchior (BBC)kicked off the final session,carefully outlining the statisti-cal analysis undertaken on theresults of an experiment look-ing at the plausibility ofbinaural synthesis with non-individualized HRTFs. Thelisteners found it difficult todetermine the headphonesimulation from the loud-speaker reproduction,although a small statisticaldifference was observed. RiittaVäänänen (Nokia) introduceda voice-guided navigationsystem for pedestrians, usinga mobile phone and anaugmented reality audio headset. In user experience testing, the devicewas found to help participants reach their destination more easily aswell as interact with the environment, device, or other people. MarkoTakanen (Aalto University) used a binaural auditory model to predictlocalization error and coloration artifacts in simulated WFS andambisonics systems, finding a good correlation with previouslycollected perceptual results. The last paper of the conference waspresented by Stephan Werner (Technical University of Ilmenau), whoinvestigated the effect of contextual factors on the perceived external-ization of binaural synthesis, suggesting that externalization isimproved by visibility of a congruent listening environment.

CONFERENCE REPORT

792 J. Audio Eng. Soc., Vol. 62, No. 11, 2014 November

Frank Melchior of the BBC presents apaper in the spatial sound psycho-physics session.

The audience listens intently during a presentation.

CONFERENCE REPORT

J. Audio Eng. Soc., Vol. 62, No. 11, 2014 November 793

POSTERSThe sixteen posters were presented over a buffet dinner on the firstevening, allowing ample time and space for discussion. Before thestart of the session, each author was asked to give a one minuteoverview of the content of their poster. There must have been somebudding soap opera writers among the authors as cliffhangers wereused to draw in the crowds.

A number of the posters focussed on binaural technologies: JesperUdesen (GN-Resound) echoed Stephan Werner’s finding that thepresence of visual cues effects externalization in binaural synthesis;Catarina Mendonça showed similar results for sound localizationwith nonindividualized HRTFs; Chengyun Zhang introduced asystem for dynamic binaural rendering of 5.1 surround sound; andTomi Huttunen and colleaguesfrom various institutionspresented a comparison of meth-ods for rapidly generating indi-vidualized HRTFs.

Crossmodel interactions werealso explored by MicheleGeronazzo, who looked at audio-haptic exploration of virtualobjects, and Samuel Moulin,who investigated audio-visualcongruence in virtual environ-ments created with WFS.

Neofytos Kaplanis (Bang &Olufsen) presented a detailedliterature review of attributesfor evaluation of reverberationin small rooms, along with asynthesized overview of thefundamental attributes andtheir subcategories.

The remaining posterscovered a wide range of differ-ent topics including micro-phone array capture andbeamforming, different soundfield rendering methods, spatialroom impulse responses, andsurround sound encoding.

WORKSHOPSThe technical program also fea-tured two workshops in theform of chaired panel discus-sions with time for questionsfrom the audience.

The first workshop was enti-tled “Spatial Audio inBroadcasting and Music: Where Are We Going?” Florian Camerer(Austrian Broadcasting) chaired the discussion, with BosseTernström (Swedish Radio), Samuli Liikanen (FinnishBroadcasting Company), and Wilfried Van Baelen (AuroTechnologies) making up the panel. Florian opened proceedingswith some facts about Finland before posing the questions “wherehave we come from?” and “where are we going?” with spatial audioin broadcasting, asking the panelists about their first experiencewith surround sound. There was generally a great deal of agree-ment between the panelists and a number of general points stoodout. Audio professionals are keen to work in new spatial formats as

technology becomes available, but it is normally challenging asthere is no increase in budget to make multiple mixes or producefor a higher number of loudspeakers. The panelists also agreed that5.1 surround is currently the most common format; whileTernström commented that 5.1 content is always preferable tostereo, Van Baelen noted that 5.1 is considerably less good thantrue 3D sound. An interesting question-and-answer sessionfollowed the culmination of the workshop, with many contribu-tions from members of the audience. Tapio Lokki started proceed-ings by asking how many members of the audience had a properlyset up 5.1 system at home; this was in the region of 10%, althoughthere was some disagreement that 5.1 did not have a wide userbase. Further questions considered binaural audio with head-track-

ing, broadcasting for car audio,and distribution over the inter-net, as well as a comment fromAngelo Farina on the currentdirection of spatial audio broad-casting in Italy.

The second workshop focussedon 3D audio quality evaluation.Andreas Silzle (Fraunhofer IIS)acted as chair, while Poppy Crum(Dolby) and Nick Zacharov(DELTA) made up the panel.Silzle started by introducing theaim of 3D audio evaluation: tocollect describing parameters ofthe audio signal processing chainin order to optimize algorithms.He went on to look at the linkbetween quality elements (thephysical domain) and qualityfeatures or attributes (the percep-tual domain). Silzle summarizedtest methodologies designed toestablish links between the twodomains, and talked briefly aboutefforts toward standardization.Nick Zacharov went into greaterdetail about the various levels ofevaluation between objective andsubjective using an updatedversion of the “filter model.” Hefurther defined the concept ofattributes, suggesting that a wordbecomes an “attribute” when itsfunction can be statisticallydemonstrated, and spoke aboutthe statistical tools available forevaluating both attributes and

assessors. Zacharov also spoke about another key part of the sensoryevaluation process: the test subjects. He suggested that an expertassessor is one who can demonstrate repeatability, discrimination, andagreement. The presentation finished with the assertion that manystudies have been performed for the purposes of attribute elicitationand called for a movement toward agreement in the community aboutthe scales that are used. Poppy Crum detailed some test methodologiesin use at Dolby, emphasizing the importance of standardized tests butalso the need to devise new methodologies where appropriate andsuggesting that it is important to correctly report experimental proce-dures rather than just falling back on the standards where they have

Delegates discuss Angelo Farina’s (center) poster, “Spatial SoundRecording with Dense Microphone Arrays.”

Florian Camerer (right) introduces the panelists in the spatial audiobroadcasting workshop, from left, Wilfried Van Baelen, SamuliLiikanen, and Bosse Ternström.

not been followed accurately. It was clear from the experience at Dolbythat expectation moderates listener judgments—this is another themethat was seen in papers throughout the conference. Again, a livelydiscussion followed the workshop with questions about agreement inattributes where listeners don’t necessarily agree on the listening expe-rience, the concept of naturalness versus plausibility, and listenerexpectation.

The two workshops left delegates with a lot to talk about duringcoffee breaks throughout the conference and think about in termsof their research and operational practice.

SOCIAL EVENTSAs well as an interesting and diverse technical program, the confer-ence organizers treated participants to two enjoyable social eventsand an informative tour of the venue.

The afternoon of the first day featured a break for outdoor activi-ties. Conference attendees were split into teams and led off to variousactivity stations, where friendly competition provided a lightheartedway for participants to get to know each other better. Activitiesincluded sheet volleyball, throwing frisbees at trees, horseshoe throw-ing, knot tying, and blow-dart shooting (the paper session chairssurely took interest in this as a potential method of stopping presen-

ters from running over their allotted time). As the events drew to aclose, the rain that had been threatening finally appeared and forced aretreat into a very small tent, where the overall winners werepresented with prizes and coffee and pastries were served. A break forexercise and fresh air was great for keeping everyone energized beforethe afternoon continued with more papers and a workshop.

The conference banquet was held on the second night atRestaurant Botta, a short walk from the conference venue.Following a tour of the Helsinki Music Centre, the banquetkicked off with a performance from Pink Twins, a Helsinki-basedduo comprising brothers Juha and Vesa Vehviläinen, whoperform audio-visual “improvised digital soundscapes.” Theperformance gave something for delegates to think about anddiscuss after Jonty Harrison’s invited lecture earlier in the day;while the music may not have been to everyone’s taste, it wascertainly an immersive and entertaining spatial audio experi-ence. Pink Twins returned for a second performance after a deli-cious three-course meal, and the socializing continued into theevening.

SUMMARYOverall, the conference was very well organized, smoothly run, and ofimmense value to all participants. The conference committee must becommended on putting together a diverse and interesting program,as well as fostering a productive environment in a desirable settingwith engaging entertainment throughout. It is clear that the area ofspatial audio is flourishing in research and in industry, and that theAES is at the forefront of technical and operational advances in thisarea. Some key points that came up throughout the conferenceincluded the improvement of timbral quality within spatial audio sys-tems, the proliferation of binaural audio and particularly individual-ized HRTFs, and the need for careful and scientific perceptual evalua-tion and modeling. Attendees were treated to numerous interestingdemonstrations and given ample opportunity for networking and dis-cussion with new and old contacts. The conference sponsors shouldbe thanked for their support: Genelec, Nokia, Neumann, Auro, andMorrow Sound (who ran a demonstration during many of the coffeebreaks), as well as Aalto University and the Sibelius Academy. TheAES will look forward to another visit to Finland in the years to come.

CONFERENCE REPORT

794 J. Audio Eng. Soc., Vol. 62, No. 11, 2014 November

Delegates get competitive with a game of sheet volleyball at theoutdoor social event.

Delegates in the concert hall during the Helsinki Music Centre tour

The conference banquet at Restaurant Botta

Editor’s note: a USB drive or downloadable PDF of the conferencepapers can be purchased online at www.aes.org/publications/conf.cfm