ViMedCSS: 24 Hours of Vietnamese Medical Code-Switching Speech

ViMedCSS provides 24.3 hours of Vietnamese medical speech across 11,832 training utterances, developed for the LREC 2026 conference. Each recording features at least one English medical term embedded within Vietnamese speech to support code-switching automatic speech recognition (ASR).

Use Cases

• Training ASR models to recognize English medical terms within Vietnamese speech
• Evaluating code-switching detection algorithms using the annotated CS terms
• Fine-tuning speech-to-text systems for Vietnamese healthcare environments

Strengths

• Contains 12,314 annotated code-switching terms in the training set
• 24.3 hours of domain-specific audio recordings
• Average utterance duration of 7.39 seconds provides consistent segment lengths

Limitations

• Small total duration of 24.3 hours compared to general ASR datasets
• High domain-specific bias toward medical vocabulary

Provenance

Source

shannonnonshan, LREC 2026

Freshness

Last updated March 2026.

Geography

Vietnam