Description of recording conditions

 

The characterisation of the signal source (the speaker), of the transducers,  and of the environment  is presented above in sections 8.3, 8.4 and 8.5. Discussion of these features will be complemented by focussing here on the signal itself, which may be described by temporal or spectral parameters. Obviously, these signal parameters have to be elucidated and supplemented by further information concerning every step of signal generation: a certain signal characteristic may be determined by the instructions that were given to the speaker before reading, by the speaker's voice characteristics , by the recording environment   and/or by any electro-acoustic component contained in the recording chain. In order to answer such questions after the recordings have been produced, information on every part of the recording chain has to be collected carefully. Refer also to Appendix D for speech recording protocols, and to Appendix C for file formats used in SAM .

The features mentioned in the following selective list may give an impression of the kind of data that have to be collected in order to ensure reproducibility:

• environment  characteristics (simple in the case of an anechoic room );
• scripting: format of the manuscript, carrier   context per recording item, additional introductory or ending texts;
• speaker setup: briefing and reading instructions, positioning of the speaker, posture of printed script (reflections!) or monitor scripting, microphone distance  and direction;
• speaking style  characteristics: speaking rate  (syllables  per second), average fundamental frequency , intensity , etc.;
• mic type and settings (if available): directional characteristics, LP-/HP-filters, use of a wind protection etc.;
• type and settings of all used electro-acoustic components (pre-amplification, analog-to-digital conversion , recording level, noise  reduction, etc.);
• overall bandwidth  by lower and upper cut-off frequency ;
• data storage media (tapes);
• digital signal  representation: sampling rate  and word length in case of linear PCM , or coding method and parameters, additional information on speech files specified by supplementary files or file headers (i.e. Windows *.wav format; refer also to Appendix C for speech file formats);
• post-processing of the speech signals: conversion of data formats, pre-emphasis, special effects, etc.