Speaker information

Speaker information should at least contain the following data:

1. Speaker/caller unique number,
2. sheet number,
3. sex,
4. age/age group/date of birth,
5. area/region (from a finite set, maximal 20 regions).

The following information is optional:

• Cordless phone,
• education level,
• size/weight of the speaker.

This information can appear once in the separate Speaker Description file which is itself referenced in each utterance label file. This avoids repeating this information in each label file. Some speaker information could be put in the label file, where we already have the row ``SPI: sex, age, language''.

It is essential that the table provides a unique mapping of information for each caller. If the speaker and sheet numbers are the same, then they are simply provided as such, but both fields must be provided. If several speakers end up calling and using the same sheet number, then there must still be some code or number which identifies that speaker uniquely and relates the other information to them.