You are here
In the CHIL project, ITC-irst collected some seminars for being used as training material. Seminars were regular presentations held from a speaker and followed by a group of listeners. Presenters used a projector and a white screen, while listeners sit in the room and had the possibility to interrupt the speaker, using their microphones.
Description of the recording audio setup
These seminars were held in the CHIL-room, a dedicated room with cameras and microphones installed on the walls. 28 microphones were mounted in 7 T-shaped arrays and then there were 4 table microphones. The speaker and three listeners wore a close talk microphone while the rest of the people shared a hand held microphone. Finally, a NIST MarkIII array modified by ITC-irst and an Acoustic Magic array were present in the room.
Data were recorded at 44.1kHz and 24-bit precision. All the channels except the MarkIII were synchronized through a clock on BNC connector. Signal format is RAW Little Endian, while for the MarkIII data format is SMD.
Table and T-array microphones: Shure MX391 boundary microphones
Closetalk microphones: Countryman E6 earset microphones
Handheld microphone: Sennheiser E845S
Description of the recording procedure
Recordings were done in an unobtrusive way. The speaker wore a radio controlled close talk microphone, in order to feel free in the movements. Another radio microphone was put on a person from the audience. Other two persons were given wired close talk microphones. A hand held microphone was put in the room, available for everybody else who wanted to ask questions. Before the beginning of the meeting the audio and video recording were started and stopped at the end.
Description of the database
ITC-irst recorded 6 regular seminars, with an audience of about 10 people everytime. Speakers were chosen trying to cover different accents: 3 Italian speakers, one American, one Dutch and one speaker coming from the German speaking part of Italy. Duration of the seminars ranged from 40 to 80 minutes.
In total there are 38 single signals from the microphones plus one MarkIII array signal for each seminar. Every file name corresponds to the microphone connected, so Array_T0-1.raw corresponds to the first microphone of the array number 0.
Recorded seminars were the following:
28/04/05, 14:30 - Luca Giulio Brayda, "Speech Recognition with microphone arrays: a careful approach to LIMABEAM algorithm"
29/04/05, 11:00 - Louis ten Bosch, "Automatic speech recognition and an application in a multimodal dialogue system"
03/05/05, 10:00 - Mauro Cettolo, "Spoken Language Translation at ITC-irst"
09/05/05, 14:30 - Massimo Zancanaro, "(Multimodal) Support for Small Group Interaction"
31/05/05, 14:00 - Oswald Lanz, "Probabilistic Multi-Person Tracking for Ambient Intelligence"
07/06/05, 14:30 - Paul Chippendale, "Collective Calibration of Active Camera Groups"