Vrije Universiteit Brussel

Home --> Research --> Nosey Elephant Studios

Nosey Elephant Studios

ETRO Audio-Visual Lab

About us

Nosey Elephant Studios is the audiovisual laboratory of ETRO based at VUB campus Etterbeek in Brussels. The lab started operation in September 2009 as a service provider in audiovisual content creation for research or industrial applications and as a platform for research valorization.

For research: ETRO based research group AVSP, active in the fields of speech/audio and video processing, audiovisual behavior analysis, speech emotion recognition and synthesis, machine vision and film/TV post-production is using this facility for capturing datasets necessary for ongoing and future research projects. The lab is offering controlled environment (acoustic, light and temperature conditions) for capturing audiovisual data using state-of-art equipment.

For industrial applications: Except working for research we can offer a number of services to externals like speech/music production, small-scale video production (with blukey), audio measurements and post-synch services for film production.

Infrastructure

AV lab consists of four rooms, a control room a recording room and two smaller rooms used for equipment storage (total area of 60 m ).

Recording Room: Is a box-in-box construction with volume of 71.2 m and internal dimensions H2.67m x W4m x L6.67m, a double window (L1m x W0.5m, 56 STC), double door (L1.8m x W0.9m, 53 STC) and a floating ceiling. The room is based on the Volkmann-Bolt design with a room ratio of 2:3:5 giving a homogeneous modal distribution. Sound pressure levels up to 100 dB-A can be produced in the room, without affecting the noise floor of neighboring offices. There is a low noise HVAC system installed (5kW heat load) keeping stable temperature and humidity. Noise floor is at 18db-A without and 22dB-A with HVAC operation. There are lines for 64 audio signals (48 inputs/ 16 sends), 6 SDI video signals and electricity. A triangular truss installed is used to hang all electricity/signal cables, lighting and cameras/microphones.

Control Room: Is a 16 m acoustically treated room with airco installed. For audio capture/processing we use a 48 channel console and a selection of pre-amps and outboard effects. There is Pro-Tools system installed with 16 channels analogue I/O and 16 channels AES/EBU I/O. For video capture we use a workstation equipped with Bluefish HD-SDI cards. Video signals are captured in raw format and all video routing is done with HD-SDI signals. The system is designed for capturing up to 6 HD-SDI and 24 audio channels. Audio and video signals are in-synch with a master clock.

Storage Room: One room covering a surface of 9 m.

Power Feed: There is a separate 3x64 Ampere feed from ETRO mains switch board. A distribution panel with ampere meter (on each phase) is installed in studio's control room. Audio and video equipment are powered from different phases.>/p>

Acoustic Treatment: AV Lab is using acoustic panels for controlling recording and control room acoustics. Design and construction of these panels was made here at ETRO. The idea was to use broadband absorption panels where depending on the wall area covered a different RT60 could be obtained. There is also a system for measuring RT60.

Research

AV Lab is providing all the infrastructure and engineering of a professional grade recording studio for the production of audiovisual databases. Multi-sensor signal processing (in-synchrony capturing and processing of multiple audio, video and human physiological signals) is an emerging research area, where data obtained through microphones, cameras and other sensors are used in order to develop, train and test new multimedia applications, enriching the human-machine interaction.

We are producing databases for the technologies being developed at ETRO, like microphone arrays, 3D surround sound, audiovisual photorealistic text-to-speech synthesis, emotional and expressive speech synthesis and recognition. In conjunction with other departments, researchers are using this facility to explore human behavior analysis, studying mother-child communication or ways of enriching the interaction between humans and robots or the gaming experience. In addition, researchers can test their algorithms using microphones, loudspeakers or video cameras under variable acoustic and lightning conditions.

Some examples of work that has been done for research:

Audio-visual database for research (2010): HD video (uncompressed 720p / 60fps) recordings with blue-key in controlled conditions. Multi-channel audio capture. Recording protocol. Database optimized for research in photorealistic audio-visual text-to-speech synthesis.

More information about the photorealistic audio-visual text-to-speech synthesis can be found here: www.etro.vub.ac.be/Personal/wmatthey/avtts/demo_AVTTS.htm

Emotional speech recordings for VUB project HOA16 (2010): Recordings of emotional non-sense speech.
Emotional audiovisual recordings for VUB project CAdE (2012): Recordings of emotional speech with facial expression.

The lab is equipped with hardware and software for basic audio/acoustic measurements. We often develop custom made measurement systems for specific needs of a project.

Some examples of work that has been done in audio/acoustic measurements:

Acoustic design (2009): Design and construction of panels for the acoustic treatment of AV Lab. Software for RT60 measurements according to ISO 3382.
Measurements for IBBT project TRACK (2010): Measurements of noise levels and RT60 in a fast moving train in the test site of Siemens in Germany.

Measurements for FP7 project ALIZE (2011): Testing of the audio and video systems of the robot NAO used in the project. Measurement and testing of localization algorithms of a 4ch microphone array and of NAO's 4ch array housed in the head, microphone array testing and calibration (ongoing work).

Measuring delays between audio and video (2011): Delays between audio and video in media production and distribution is one of the most crucial maters. We developed a prototype system for measuring the true delay in the recordings between the audio and video signal, captured by a camera and a microphone. Our measurement has an accuracy of +/- 8msec. This system can be used when audio and video are recorded independently without a common time reference.

Services

AV Lab is offering to VUB community and external costumers services in speech, music and video production. Services are provided in the studio or on location. Some samples of our work can be found here:

Video recording of the speech of philosopher Isabelle Stangers at VUB: