What languages does Medical Mode support for medical transcription?

English, Spanish, German, and French, on both pre-recorded and realtime audio.

Can it handle a visit where the speaker switches languages?

Yes — Universal-3.5 Pro supports native intra-utterance code-switching, so a bilingual conversation transcribes as one coherent record.

Is the medical accuracy the same in Spanish as in English?

Medical Mode applies the same clinical vocabulary tuning in all four supported languages. Published Missed Entity Rate figures are not broken out per language, so validate on your own audio in the playground.

Do I need a different model per language?

No. One model, one parameter (domain: "medical-v1") covers all four languages.

What about medical transcription in languages beyond these four?

For broad coverage without Medical Mode, Universal-2 supports 99+ languages; Medical Mode itself is EN/ES/DE/FR today.

How much does multilingual medical transcription cost?

$0.15/hr add-on, the same price in every region — including EU data residency via api.eu.assemblyai.com, where audio and transcripts stay in the EU.

Multilingual medical

Clinical accuracy in English, Spanish, German, and French

Multilingual medical transcription captures clinical encounters in English, Spanish, German, and French, including mid-sentence code-switching. AssemblyAI's Medical Mode, a one-parameter add-on to Universal-3.5 Pro, holds a 3.2% Missed Entity Rate on clinical conversations and runs in every supported language on both pre-recorded and streaming audio. BAA available.

Real clinical conversations switch languages mid-sentence. Medical Mode runs on Universal-3.5 Pro across four languages with native code-switching, so a Spanish-English visit transcribes as one accurate record — not two broken ones. One parameter, async or realtime.

Try Medical Mode free on your audio Read the docs

Medical entity accuracy that holds across languages

Missed Entity Rate 3.2%

Missed Entity Rate on clinical conversations with Medical Mode.

Fewer missed entities ~20%

Fewer missed medical entities than the base model without Medical Mode.

Fewer entity errors 87%

Fewer medical entity errors than the base model.

Benchmarked providers #1

Lowest Missed Entity Rate vs. Deepgram, Speechmatics, AWS, and Google.

Benchmarks

Multilingual medical transcription accuracy vs. other providers

Missed Entity Rate is the share of clinical entities — medication names, dosages, diagnoses — that a model fails to transcribe correctly. Lower is better.

Missed Entity Rate and Word Error Rate on clinical conversation audio Lower is better. Medical Mode enabled via `domain: "medical-v1"`.
Provider	Missed Entity Rate	Word Error Rate
AssemblyAI Universal-3.5 Pro w/ Medical Mode	3.2%	5.3%
Speechmatics Enhanced Medical	3.6%	5.5%
Deepgram Nova-3 Medical	4.7%	6.1%
AWS Transcribe Medical	8.7%	5.9%
Google Medical Conversation	24.4%	12.9%

See the full benchmarks

Activation

One parameter, four languages

No model swap — one parameter covers EN, ES, DE, and FR, on async or realtime.

import assemblyai as aai

aai.settings.api_key = "YOUR_API_KEY"

# Multilingual medical: ES/EN code-switching clinical visit
config = aai.TranscriptionConfig(
    speech_models=["universal-3-5-pro", "universal-2"],
    domain="medical-v1",
    keyterms_prompt=["hipertension", "lisinopril", "diabetes tipo 2"],
)

transcript = aai.Transcriber().transcribe("./visita-clinica.wav", config)
print(transcript.text)

Four clinical languages

EN, ES, DE, FR, pre-recorded and realtime, all with Medical Mode.

Medical Mode

Native code-switching

Handles intra-utterance language switches without splitting the session.

Universal-3.5 Pro

Private by default

PHI redaction across audio and transcripts, BAA available, SOC 2 Type 2.