CoRal V2: Danish Conversational and Read-Aloud Speech with 100K+ Records

CoRal V2 provides between 100,000 and 1,000,000 audio records of Danish speech for Automatic Speech Recognition (ASR) tasks. Created by the CoRal-project and updated in June 2025, the collection includes both conversational and read-aloud samples across diverse dialects, accents, and age groups.

Use Cases

• Training ASR models to handle regional Danish dialects and accents
• Classifying audio samples based on speaker age groups or gender labels
• Transcribing spontaneous conversational Danish speech versus scripted read-aloud text

Strengths

• Scale of 100,000 to 1,000,000 records
• Dual-modality coverage of both conversational and read-aloud speech
• Inclusion of diverse Danish accents and regional dialects

Limitations

• Restricted to the Danish language, limiting utility for multilingual applications
• Metadata regarding specific recording environments or hardware is not provided

Provenance

Source

CoRal-project

Collection Method

Recorded conversational and read-aloud speech

Freshness

Updated as of June 2025.

Geography

Denmark