Spanish Medical Dialogue Dataset (TEME-v1)

This dataset contains 90 Spanish medical dialogues across multiple clinical specialties, including cardiology, primary care, and pulmonology. These dialogues have been obtained from credible cases and reviewed by a medical professional.Each dialogue includes:

The dialogue (json) of the conversation between doctor and patient.

The corresponding audio recording (.mp3) of the same dialogue.

The corpus was designed to reflect realistic clinical encounters while ensuring complete anonymization of patient information.It serves as a benchmark for medical speech recognition and automatic transcription evaluation in Spanish.

Distribution across specialties:• Cardiology: 30• Primary Care: 17• Pulmonology: 8• Neurology: 8 • Traumatology: 5• Endocrinology: 5• Oncology: 5• Geriatrics: 4• Emergency: 4• Pediatrics: 4

The fields used in the json for each file are described below:

scenario_id: random identifier for the scenario.

title: title related to dialogue.

context: brief summary of the dialogue

turns: doctor-patient medical dialogue with its corresponding content.

speaker: custom name used as identifier in the synthesizer, indicating the person who is participating at that time.

text: content of the dialogue.

duration_estimate: time between shifts to indicate the synthesizer.

voice_characteristics: voice characteristics to indicate the synthesizer.