Three types of speakers participated in the recordings:
- Professional radio speakers (4 per language)
- Professional advertising/dubbing speakers (4 per language)
- Non-professional speakers (20 per language)
They are organized into 4 categories, depending on the amount of material recorded:
- A: Prosodic news corpus + Task oriented dialogue corpus + Informal dialogue corpus
- B: Prosodic news corpus + Prosodic news corpus
- C: Task oriented dialogue corpus + Informal dialogue corpus
- D: Task oriented dialogue corpus
Each speaker has a unique Glissando ID, which includes:
- a letter indicating the gender: female ('f') or male ('m')
- a two-figure number, unique for each speaker in the corpus
- a letter indicating if the speaker is radio professional ('r'), advertising/dubbing professional ('a') or non-professional ('s')
Examples:
- f01r: speaker 1, female, radio professional
- m05a: speaker 5, male, advertising/dubbing professional
- f37s: speaker 37, female, non-professional
They have also a three-letter short ID:
- f01r: ANN
- m05a: JVA
- f37s: MIRl