Multilingual Videoconferencing Solutions Benchmark Initial Findings Workshop Prepared for Directorate General of Interpretation Guido van der Harst Managing Partner Arjen Ringers Associate Director Gartner is a registered trademark of Gartner, Inc. or its affiliates. 0
Agenda Introduction Technical preparation of the meeting Participant preparation of the meeting During the meeting Next steps Gartner is a registered trademark of Gartner, Inc. or its affiliates. 1
Objectives of today s meeting Share with all stakeholders The findings from the interviews Analyses of these findings Initial recommendations and suggestions Assure all aspects are being included Gartner is a registered trademark of Gartner, Inc. or its affiliates. 2
Multiple institutions have provided input in the analyses More than 25 persons have been interviewed The following institutions have provided input through interviews or provided documentation: EC (COMM, DIGIT, SANCO, SCIC) Consilium Court of Justice EUROJUST European Economic and Social Committee IMF Organisation for Economic Co-operation and Development The maturity levels of the interviewed institutions differ: Technology such as equipment, network, standards Process manuals that exists to support participants, well established procedures Gartner is a registered trademark of Gartner, Inc. or its affiliates. 3
The scope is still somewhat challenged Type of meeting different kind of meetings will require different approaches or emphasis on aspects: Court meetings (truth th for live) fixed lay-out, very regulated on both procedures and camera and participant i t positioning Negotiations, often no agenda or facilitator, less structured and more interactive Normal meetings (our scope) structure, meeting chairman Although the outcome of this project is anticipated to be useful for all institutions and meeting types, it s scope is limited to normal meetings with simultaneous interpretation Technical aspects are out of scope as the previous Gartner study and Fraunhofer study have set the boundaries regarding the minimum i perceived quality and technical requirements for meetings with video conference intervention and simultaneous interpretation However, Gartner will discuss sound/microphone aspects in slightly more (technical) detail, as not all aspects of this subject were discussed in detail during the previous study The scope was often challenged with the question why mobile devises or executive/personal desktop solution where out of scope Gartner is a registered trademark of Gartner, Inc. or its affiliates. 4
Gartner considers the immersive and Room-based VC solutions as viable alternatives for a distant site; the personal desktop is a viable alternative for Use Case 1 only when appropriate equipment is used VC end-points Use Case 1 Use Case 2 Use Case 3 Immersive (telepresence) Room-based Executive Desktop Personal Desktop Mobile Current scope Low Level of viability High Not applicable at all Very significant customization effort required Significant customization effort required Average customization effort required No or limited customization effort required Gartner is a registered trademark of Gartner, Inc. or its affiliates. 5 Note: Based on Interpreter requirements For the distant sites, UC2 and UC3 are similar
Only fully converged networks and specifically dedicated video overlay networks are capable of delivering the bandwidth required for simultaneous interpretation via videoconferencing Network performance has greatest influence on perceived quality of the end-to-end VC solution Network Capability for VC Comments Wireless, satellite, ISDN Does not provide sufficient bandwidth or quality Internet Only effective if delivered directly to the site and needs to be from the same ISP and delivered in-region for sufficient performance Not recommended as a primary network Fully converged network (Combining VC traffic with existing private IP networks) Dependent on quality and bandwidth of existing networks Very complex Requires experience in managing real-time traffic Overlay network Highly scalable (VC traffic on separate IP Expensive networks) Overlay networks can be provided by third parties Not capable Capable in very limited cases Capable in some cases Capable in majority of cases Capable in almost all cases Gartner is a registered trademark of Gartner, Inc. or its affiliates. 6
Agenda Introduction Technical preparation of the meeting Camera Sound Light Room lay-out Participant preparation of the meeting During the meeting Next steps Gartner is a registered trademark of Gartner, Inc. or its affiliates. 7
Camera overall observations Gaze angle The most disturbing situation with regards to camera positioning is when the distant party does not look into the camera and as such it is not looking at the other party, speaking into the void Horizontal zooming In some cases the camera stays completely zoomed out and shows one participant several empty chairs. The face of the participant is hardly visible Vertical zooming Although there are different opinions, the majority of the interviewees like to see the distant participant from the waist up including the hands, the so called news reader position Horizontal angle The horizontal angle is sometimes to large, resulting in a picture of the participant from the side where the lips are hardly visible Vertical angle The vertical angle has not been mentioned often as a problem and is in most situations correct White balance The picture looks unnatural, people look red or green Gartner is a registered trademark of Gartner, Inc. or its affiliates. 8
Overall positioning the most important aspect of video conferencing is looking into the camera at your audience; reduced the gaze angle Communicating with you Gaze angle refers to the angle between your camera centre-field and your own line of sight. The difference is called the gaze angle If this Angle is too high participants look into the void and Have no eye contact with the receiver A high gaze angle looks unnatural The gaze angle can be reduced by putting the camera right next or above the monitor The gaze angle becomes less important when the camera distance is higher Communicating with the PC Gartner is a registered trademark of Gartner, Inc. or its affiliates. 9
A horizontal camera angle between 0 and 45 degrees is acceptable, beyond 45 degrees lips are hardly noticeable. Let the participant directly face the camera if impossible 90 degree 45 degree 0 degree A horizontal angle deviation might not always be the cause of the camera Especially with large rooms, it might not be possible to have a camera positioned in such a way that the speaker looks at the screen, computer and camera at the same time The speaker should be made aware of where the camera is Gartner is a registered trademark of Gartner, Inc. or its affiliates. 10
Horizontal angle should be as straight as possible Table mounted Monitor mounted Ceiling mounted Distorts proportions p of the face Most ideal position Distorts proportions p of the face Gartner is a registered trademark of Gartner, Inc. or its affiliates. 11
The most preferred camera position for VC with simultaneous interpretation is the news reader position where face and arms are clearly visible News reader Waist up Talking head Most preferred position for interpreters Difficult to observe body language Easy to get partly out of the picture The head will quickly move out of the frame Lighting will be more difficult Gartner is a registered trademark of Gartner, Inc. or its affiliates. 12
Sound overall observations, including the use of the Microphone Talking direction Participants often don t speak in the direction of the microphone or tend to steer away. As such the sound becomes more difficult to hear Obstructed t sound Participants i t put a laptop or a stack of documents before the microphone. The sound gets even worse Ambient noise Background (static) noise is picked up by the microphone, also if the distant party is not speaking Missing frequencies In some cases where the distant party has the proper microphone setup, it may seem that certain frequencies are missing; the microphone does not cover the full audio frequency spectrum Resonance or reverberation Lots of resonance and echoes. Participants sound as if they are located in the bathroom Behavioural aspects will be discussed in the next chapter Noise caused by shifting papers around or ticking on the table Testing the microphone by blowing into it or tapping on it Gartner is a registered trademark of Gartner, Inc. or its affiliates. 13
Testing several microphones, obstructing sound and ambient noise The video has been shortened but the content of the video has not been changed Source: PCM Marketing official dealer of Shure in California For the full film see: https://www.youtube.com/watch?v=8-gityt4eac Gartner is a registered trademark of Gartner, Inc. or its affiliates. 14
Although a shotgun microphone (or hypercardioid ) microphone or headset might seem a good solution, they introduce additional challenges Shotgun microphone The sound of a shotgun microphone can sometimes be unnatural and It is very difficult to keep the sound consistent Shotguns can have an area of increased sensitivity directly to the rear Directional microphones work good especially when placed on a gooseneck Headset microphones are generally unidirectional (like the shotgun) with a much smaller field of pickup but It blocks a big part of the face Wires get in the way More difficult to wear with glasses Omnidirectional microphones generally pick up a lot of ambient noise Directional microphone Omnidirectional microphone Gartner is a registered trademark of Gartner, Inc. or its affiliates. 15
Having the right microphone does also depend on the frequency response curve It s not the frequency range of the microphone that matters but the frequency response pattern A microphone's frequency enc response pattern relates the frequency enc in Hertz (x axis) to the response in decibels (y axis) In many cases a tailored frequency response is more useful than a flat one. As a general rule: If the microphone is suitable for speech recognition, it is suitable for simultaneous interpretation Low frequency is less amplified Gartner is a registered trademark of Gartner, Inc. or its affiliates. 16
To achieve quality audio, it's important to reduce reverberation (resonance) The video has been shortened but the Content of the video has not been changed Source: Acoustic Geometry For the full film see: https://www.youtube.com/watch?v=jpyt10zrclq Gartner is a registered trademark of Gartner, Inc. or its affiliates. 17
Although proper lighting is not easy to achieve, it is one of the most important ingredients for a sharp picture; natural light can be very disruptive Light from the back Windows do not only distract, but also provide disruptive lighting Light from the side Gartner is a registered trademark of Gartner, Inc. or its affiliates. 18
Although proper lighting is not easy to achieve, it is one of the most important ingredients for a sharp picture; artificial light needs to be placed correctly Light from the side Light that is only projected from a certain angle will cause hard shadows Light from the top Gartner is a registered trademark of Gartner, Inc. or its affiliates. 19
Lighting Challenges with light Using direct lighting, Not using indirect diffused light Mixing light types (different colour temperatures ) No fill light light that is used to eliminate shadows mostly top down No background light to evenly light the back walls (most difficult one) There is still some shadow (see chin, right arm and eyes) but not disturbing Note Shadows are not necessarily bad, as long as they are not too hard. No shadows at all, results in a very unnatural picture Gartner is a registered trademark of Gartner, Inc. or its affiliates. 20
Room design overall observations Fatigue Some interviewees claimed that some distant locations caused them to be more tired than other distant locations, although the sound was good Poor colours During some video conferences, people use a flag or logo s at the back, not providing a clear picture of the participant Other objects Some conference rooms are clearly more suitable for meetings then for video conferences. In such rooms there are lots of objects visible such as water bottles, lunch trolley's etc. This is perceived as very distractive Room is too large In some cases a large room is used without the camera zooming in on the participant. i t This results in a very small picture of the participant i t who s face is hardly visible Gartner is a registered trademark of Gartner, Inc. or its affiliates. 21
Agenda Introduction Technical preparation of the meeting Participant preparation a of the meeting Clothing Presentation slides Support During the meeting Next steps Gartner is a registered trademark of Gartner, Inc. or its affiliates. 22
Clothing, presentation slides and support; overall observations Clothing Most interviewees had difficulty with clothing that had busy prints or stripes. These often appear blurred (all technical people referred to stripes in clothing as something to avoid) Switching between conference room view and slides Interviewees had difficulty focusing on the distant participants when frequent switching between the presentation and the participants occurs. This is caused by brightness and contrast differences Missing support There were multiple cases where bad sound quality could have easily been solved if support at the distant site would have been available. This is an observation that also refers to the section technical preparation and during the meeting Desk discipline In some cases there were a lot of objects on the desk, distracting the viewers and creating unwanted noise Self view This is not always turned-on. As such, participants do not realize they are not properly in the picture or lighted. Some people feel uncomfortable seeing themselves or have a small monitor and do not want to consume space or block part of the remote image Gartner is a registered trademark of Gartner, Inc. or its affiliates. 23
Clothing it is not only the print or striping that matters, but also the contrast within the print or striping Although both types of clothing are not preferred, contrast and colours make a difference in the amount of focus that is needed to create an accurate picture and the amount of data that needs to be transmitted Resolution; 1944x2592 Size; 1,89 MB Resolution; 1944x2592 Size; 544 KB Gartner is a registered trademark of Gartner, Inc. or its affiliates. 24
Clothing With a low resolution camera, the effect of the print or striping results in high blurring of the picture The combination of striping and movement of a person has a significant impact on the amount of data that needs to be transmitted Resolution; 640x480 Size; 54 kb Resolution; 1944x2592 Size; 544 KB VGA resolution HD resolution Gartner is a registered trademark of Gartner, Inc. or its affiliates. 25
Clothing does not only affect the amount of data that will be transmitted but also the amount of light that is needed to clearly capture the picture 71 % reduction 31 % reduction 14 % reduction Resolution; 1944x2592 Size; 1,89 MB ¼ Shutter speed Resolution; 1944x2592 Resolution; 1944x2592 Size; 544 KB Size; 376 KB 1/6 Shutter speed 1/6 Shutter speed Resolution; 1944x2592 Size; 326 KB 1/8 Shutter speed Shutter speed reduces (less light is needed to capture the image with same aperture) Gartner is a registered trademark of Gartner, Inc. or its affiliates. 26
EXAMPLE SLIDE 1 Multilingual Videoconferencing Solutions Benchmark Initial finding workshop Prepared for Guido van der Harst Managing Partner Arjen Ringers Associate Director Gartner is a registered trademark of Gartner, Inc. or its affiliates. 27
EXAMPLE SLIDE 2 Gartner is a registered trademark of Gartner, Inc. or its affiliates. 28
Switching between slides and the conference room view can cause fatigue at participants ISO 100 ISO 100 ISO 100 Exposure time1/30 Exposure time1/30 Exposure time1/30 F-stop f/6,3 F-stop f/3,5 F-stop f/3,2 Examples of Aperture The aperture of a camera works the same as the human eye; with more light the iris tightens and with less light it opens Frequent switching between slides and conference room view can therefore put a strain on the eyes of the viewer Gartner is a registered trademark of Gartner, Inc. or its affiliates. 29
Desk discipline although the amount of items in the picture might not affect the data transfer, it distracts from the main subject and can create additional noise Lots of objects on the desk will disturb the viewer Hands become less visible The sound path can be obstructed Light reflects differently. This may also have an impact on the white balance It is more likely that disturbing sounds, such as the moving of paper, will be picked up by the microphone Lots of objects will not necessarily have a great impact on the data transfer as they do not move and as such do not need to be refreshed Gartner is a registered trademark of Gartner, Inc. or its affiliates. 30
Agenda Introduction Technical preparation of the meeting Participant preparation of the meeting During the meeting Next steps Gartner is a registered trademark of Gartner, Inc. or its affiliates. 31
During the meeting Overall observations People not looking into the camera This has been the biggest issue for most interviewees Testing the microphone Additional background noise Background noise can range from street noise coming from an open window, office noise or noise in the room Dynamic backgrounds In rare occasions there are screens with screensavers on the background or part of an office (behind a glass wall) Talking into the microphone In some cases they are too far and in some cases too close. Talking in another direction than where the microphone is positioned also results in lower audio quality Gartner is a registered trademark of Gartner, Inc. or its affiliates. 32
During the meeting Overall observations (continued) Shut down the microphone many participants do not mute the microphone when they are not speaking Movement depending on the video quality settings and available bandwidth, gestures may appear "jumpy. The picture becomes unclear when participants move a lot and fast Sitting position participants tend to sit at the end of a table in cases where there is no zoom functionality. This results in a very unbalanced view Self view This is often not used properly. As such participants are not aware of how they are viewed by the distant t site On time participants do not always start on time. Audio cannot be tested. Guidance to the distant location if sound is not properly captured is not possible Gartner is a registered trademark of Gartner, Inc. or its affiliates. 33
Agenda Introduction Technical preparation of the meeting Participant preparation of the meeting During the meeting Next steps Gartner is a registered trademark of Gartner, Inc. or its affiliates. 34
Next steps If you have any additional input or comments please share them at arjen.ringers@gartner.com Additional input will still be included as far as applicable and if submitted before 25 th of March 2014 The 29 th of April 2014 the final presentation of the file will be held in the Centre de Borschette room 5B from 10 o'clock., Gartner is a registered trademark of Gartner, Inc. or its affiliates. 35
Contacts Client Contact t Gartner Contact t Lode De Raedt Arjen Ringers Associate Director Gartner Consulting arjen.ringers@gartner.com Guido van der Harst Managing Partner Gartner Consulting guido.vanderharst@gartner.com Gartner is a registered trademark of Gartner, Inc. or its affiliates. 36