Danish ASR Leaderboard: Benchmark Results Across Five Test Sets

Five independent public test sets evaluate Danish automatic speech recognition models. Each row represents one evaluated model, with scores reported as Word Error Rate (WER) and Character Error Rate (CER) percentages. The dataset, created by RyeAI, was last updated on 2026-06-19.

Use Cases

• Benchmark model performance based on Word Error Rate and Character Error Rate scores
• Compare ASR models across different domains like spontaneous conversation and read speech
• Identify model strengths and weaknesses using results from five independent test sets

Strengths

• Results are derived from five independent public test sets, including CoRal-v3 for conversation
• Evaluation is described as reproducible and open, supporting transparency
• Scores are provided as both Word Error Rate (WER) and Character Error Rate (CER)

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment
• Last updated 2026-06-19 08:33:32; freshness should be verified

Provenance

Source

huggingface, author RyeAI

Collection Method

Benchmark results backing the Open Danish ASR Leaderboard

Freshness

Last updated 2026-06-19 08:33:32

Geography

Denmark (Danish language focus)