What is the best speech-to-text API for medical dictation?

For dictation, accuracy on drugs, dosages, and abbreviations matters most. Universal-3.5 Pro with Medical Mode reaches a 3.2% Missed Entity Rate — the lowest across benchmarked providers — in real time or batch.

Can clinicians dictate medical notes in real time?

Yes, on Universal-3.5 Pro Realtime at ~150ms P50 latency, with text appearing as they speak.

How do I handle specialty-specific medical vocabulary?

Use keyterms_prompt (up to 1,000 terms async, 100 realtime, updatable mid-stream) to bias toward your specialty.

Is dictated PHI protected and BAA-eligible?

AssemblyAI is a business associate under HIPAA and signs a standard BAA; PHI redaction; SOC 2 Type 2.

How much does real-time medical dictation cost?

$0.15/hr Medical Mode add-on on top of the model rate.

How do I enable Medical Mode for clinical dictation?

Add domain: "medical-v1" — no model swap.

Clinical dictation

Dictation that gets the medicine right the first time

Clinicians dictate fast, with drug names, dosages, and abbreviations packed into every line. Universal-3.5 Pro with Medical Mode posts a 3.2% Missed Entity Rate, so dictated notes need fewer corrections and round-trip less. Real-time as they speak, or batched from recordings.

Try Medical Mode free on your audio Read the docs

Fewer missed entities means fewer dictation corrections

Missed Entity Rate 3.2%

Missed Entity Rate on clinical conversations with Medical Mode.

vs. Universal-3 Pro ~20%

Fewer missed medical entities than Universal-3 Pro alone.

vs. base model 87%

Fewer medical entity errors than the base model.

Benchmarked providers #1

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

See the medical benchmarks

Activation

Stream dictation straight into your note

Stream dictation into the field as the clinician speaks — one WebSocket, one parameter, no model swap.

from assemblyai.streaming.v3 import StreamingClient, StreamingParameters

client = StreamingClient(api_key="YOUR_API_KEY")

# Clinical dictation: fast speech, dense terminology
client.connect(StreamingParameters(
    speech_model="universal-3-5-pro",
    domain="medical-v1",
    keyterms_prompt=["q.d.", "p.r.n.", "amoxicillin", "500 mg", "BID"],
))

Dense terminology, handled

Abbreviations, dosing shorthand, and drug names captured so dictated text needs less cleanup.

Universal-3.5 Pro

Real-time entry

~150ms P50 streams dictation into the field as the clinician speaks.

Universal-3.5 Pro Realtime

Tune to a specialty

Pass up to 1,000 keyterms (async) to bias toward a specialty's drugs and procedures.

Medical Mode

Common questions

: For dictation, accuracy on drugs, dosages, and abbreviations matters most. Universal-3.5 Pro with Medical Mode reaches a 3.2% Missed Entity Rate — the lowest across benchmarked providers — in real time or batch.
: Yes, on Universal-3.5 Pro Realtime at ~150ms P50 latency, with text appearing as they speak.
: Use keyterms_prompt (up to 1,000 terms async, 100 realtime, updatable mid-stream) to bias toward your specialty.
: AssemblyAI is a business associate under HIPAA and signs a standard BAA; PHI redaction; SOC 2 Type 2.
: $0.15/hr Medical Mode add-on on top of the model rate.
: Add domain: "medical-v1" — no model swap.

Try Medical Mode free on your audio

Run your own dictation through Medical Mode in the playground — one parameter, real time. Free to start, no credit card required.

Try Medical Mode free