next up previous contents
Next: Speaker Profiles Up: Meta Data of SpeechDat Previous: Meta Data of SpeechDat   Contents

Recording Protocol

The SpeechDat-II partners were free to choose their recording equipment. Most partners used the ADA software developed at the Polytechnical University of Catalonia in Barcelona, or proprietary software. For the German data collection, proprietary software was used. The recording platform was a standard PC running Windows NT connected to an ISDN prime rate interface (30 channels). Some recording meta data was registered directly by the recording software, e.g. date and time of call, session ID, etc., other data was added later during the preparation of a recording session for transcription, e.g. prompt sheet number, region and network of call, etc.

The following list of minimal requirements for the recording protocol as given in section [*] contain either a + or - if the content is given or a citation from the documentation files.





Session ID +
Speaker ID +
Date of recording +
Environmental conditions Room Acoustics: + class of 3 different environments
Sources of Noise/Background Noise: -
Cross Talk: -
Technical recording conditions (in documentation)
Microphones (in documentation)
Recording device (in documentation)
Technical specifications of recorded signals (in documentation)
Placement and distance to microphone(s) -
Name or ID of the recording +
Details about the recorded domain(s) n.a.
Details about instruction to speaker(s) -
Duration of the recording session +
Type of Prompting (paper, face-to-face, screen, voice) paper (in documentation)
Emotional speech yes/no -
Details about acoustics (reverberation, S/N ratio etc.) -
Supervisor present n.a.
Interpreter present n.a.
WOZ: details about `virtual machine' n.a.
Type of speech read (in documentation)
Free comments -


next up previous contents
Next: Speaker Profiles Up: Meta Data of SpeechDat Previous: Meta Data of SpeechDat   Contents
BITS Projekt-Account 2004-06-01