Add the complete database

[felixbur]

The complete data described in the paper has in addition

• also the files that were not recognized by 80% of the listeners
• the laryngogram tracks

[felixbur]

ok, so i would actually start doing this now. I got here 817 stereo files (audio and laryngo channels).

I would

• split the channels
• add the files that are not already in the distribution to files table
• add the additional files to a table named emotion.categories.extra.gold_standard
• make an extra folder for the laryngo tracks
• multiply all tables with the prefix "laryngo", eg, "laryngo.files", "laryngo.emotion.categories.extra.gold_standard" etc.

what do you think, @hagenw ?

[hagenw]

split the channels

Why do they have more channels. Stereo recordings, or different speaker in different channels?

make an extra folder for the laryngo tracks

What is the file format of the laryngo tracks? I guess you would like to tread them as media files and create tables having them in the index, correct?

[felixbur]

1) left channel= audio, right channel laryngo 2) all in wav 16kHz format

[felixbur]

btw i see that there also is a naturalness value ("how natural sounds the sample?").

I would add this as well

[schruefer]

Cool! I'm looking forward to using those additional tracks!

add the additional files to a table named emotion.categories.extra.gold_standard

I would not call this table gold_standard since there is no agreement, which one would expect by this name. Rather just call it emotion.categories.extra or emotion.categories.disagreement or emotion.categories.ambiguous

[felixbur]

emotion.categories.ambiguous seems best