Timbuktu
Data catalogue

CONV-001

Illustrative dataset format

Multilingual Customer Conversations

Natural two-party service conversations across several languages, presented as an illustrative dataset format.

Country / region
Multi-market · Global
Availability
Evaluation
Modality
Audio
Languages
English · Spanish · French

Technical characteristics

Two-party conversations with partial transcripts and speaker metadata.

Potential applications

Model training, adaptation and evaluation use cases relevant to conversational systems, subject to the final dataset specification and licence.

Rights & provenance

Consent and licensing scope would be verified before access.