Back to homepage

Jasmin-CGN

• Results
• Setup

Corpus description

Jasmin-CGN is a Dutch/Flemish corpus that contains speech from less-represented groups, such as the elderly, children, or non-natives.

The corpus is split mainly according to 2 criteria:

1. Dutch/Flemish
2. Read/HMI speech
3. Speaker groups based on age and native/non-native

Thus, we have:

• comp_p: Speech recordings where a human interacts with a machine (HMI) in a Wizard of Oz setup
• comp_q: Speech recordings of people reading a text

And the speaker groups:

1. Native children (ages 7-11)
2. Native teenagers (ages 12-16)
3. Non-native children (ages 7-16)
4. Non-native adults (ages 18-60)
5. Native elderly (ages 65+)

For more details about the corpus, check the documentation. The corpus, including its documentation, can be downloaded from here. The paper released for the corpus can be found here.