Med-Dictate: Medical Speech Recognition Benchmark in Three Languages

Med-Dictate is an evaluation dataset released by Corti ApS alongside the Symphony for Speech Recognition white-paper. It contains medical notes dictated by Corti team members and one contractor, with their written consent, in English, French, and German. The dataset is built for benchmarking automatic speech recognition and related NLP systems on medical-domain audio.

Use Cases

• Benchmarking ASR system performance based on medical-domain audio
• Evaluating NLP systems on transcribed medical notes
• Training or fine-tuning multilingual speech models based on the English, French, and German content
• Studying domain adaptation for speech recognition based on the clinical context

Strengths

• Contains audio and text in three languages: English, French, and German
• Explicitly built for benchmarking ASR and NLP systems
• Uses consented, non-patient data with no PHI or identifiable third-party content

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment
• Data may reflect source bias inherent to being dictated by a limited team from Corti ApS

Provenance

Source

Corti ApS

Collection Method

Medical notes dictated by Corti team members and one contractor with written consent.

Freshness

Last updated 2026-06-10 07:43:47; freshness should be verified