Top Doctors' AI medical scribe for clinical reports - Voice AI Space Barcelona
Explore how AI medical scribes automate clinical documentation, reducing administrative workloads while improving patient care quality and reporting efficiency today.
Summary
Julien Delonay, an AI researcher at Top Doctors, presented TD Listener, a medical AI scribe tool designed to reduce the administrative burden on healthcare professionals. By automating the creation of clinical reports, the tool allows doctors to focus more on patient interaction rather than typing notes during consultations.
Key Features and Functionality
- Audio Input Options: Doctors can record consultations using a computer microphone, a mobile phone (via a QR code), uploaded audio files, or by integrating the tool directly into telemedicine platforms like Microsoft Teams, Zoom, and Google Meet.
- Customizable Templates: The platform supports various report formats, such as first visits or follow-ups, and allows doctors to easily create custom templates.
- SOAP Structure: Reports are generated following the standard SOAP (Subjective, Objective, Assessment, Plan) format.
- Multilingual Support: The tool currently supports English, Spanish, Italian, and Catalan.
- Error Prevention: To address potential AI hallucinations, the system highlights critical terms like medications and dosages in brackets, prompting doctors to review and edit them before saving.
Model Evaluation and Research
The development team evaluated several automatic speech recognition (ASR) models—including Whisper, GPT Transcribe, and Google Cloud Speech-to-Text—alongside various large language models (LLMs) like GPT, Gemini, and Med-Gemini. Their research indicated that general-purpose models performed better and were more cost-effective for generating accurate clinical reports than specialized medical models.
Data Privacy, Pricing, and Integration
- Data Security: To comply with data protection regulations, Top Doctors utilizes Microsoft Azure under an agreement that prevents patient data and audio from being used for model training. Audio files are not stored long-term.
- Pricing: Pricing varies by region; for example, in Spain, the service costs $30 for 60 hours of audio.
- System Integration: TD Listener is available to the 90,000 doctors within the Top Doctors network and can be integrated into hospital and clinic software systems.
Future Developments
Planned features include automated scheduling integration, which will allow the system to identify open calendar slots and send follow-up appointment invitations to patients via WhatsApp. Additionally, the team aims to implement a feature that generates brief, three-to-four-sentence patient summaries for doctors to review prior to consultations.
Related Content
Gradium's on-device, CPU-only text-to-speech for private voice AI - Voice AI Space Barcelona

Demo - ChickyTutor, an AI language tutor for everyone - Voice AI Space Amsterdam

Agora's real-time network powers conversational AI and avatars - Voice AI Space Barcelona

Enera's Voice AI agent for EV charger support - Voice AI Space Barcelona

Palabra AI launches fast real-time translation and TTS - Voice AI Space Barcelona

Detecting sarcasm with Voice AI - Voice AI Space Amsterdam

Vibe Coding a Voice AI Agent with Claude & Vapi - Voice AI Space Amsterdam

Lessons learnt from building an AI Voice Assistant for scientific labs - Voice AI Space Amsterdam