Can AssemblyAI output structured data, not just a transcript?

Yes — entity detection (50+ types) runs alongside transcription, giving mappable entities for EHR fields and coding.

Does it redact PHI from clinical transcripts?

Yes — PHI redaction across both audio and transcripts, in the same request.

Can we deploy it in our own VPC or EU region?

Yes — Voice AI Cloud, EU data residency (api.eu.assemblyai.com), or self-hosted in your AWS VPC.

How do we validate accuracy on our data?

Run your own audio against the benchmarks via Applied AI — we'll measure MER on your real encounters.

What compliance posture do you offer?

Business associate under HIPAA with a standard BAA, PHI redaction, and SOC 2 Type 2.

How is it priced at scale?

Pay-as-you-go per second, no minimums; Medical Mode is a $0.15/hr add-on.

EHR & RCM platforms

Speech-to-text that feeds structured clinical data

EHR and revenue-cycle platforms need more than a transcript — they need accurate entities to map to codes, fields, and claims. Universal-3.5 Pro with Medical Mode delivers a 3.2% Missed Entity Rate plus entity detection and PHI redaction, so what flows into your system is clean at the source.

Run benchmarks on your own data Read the docs

Accurate entities in means accurate codes and claims out

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

Transcript + entities + redaction in one pass

One request returns the transcript, detected entities, and redacted PHI — ready to map to your EHR fields and claims.

import assemblyai as aai

aai.settings.api_key = "YOUR_API_KEY"

# EHR/RCM: medical transcript + entity detection + PHI redaction
config = aai.TranscriptionConfig(
    speech_models=["universal-3-5-pro", "universal-2"],
    domain="medical-v1",
    entity_detection=True,
    redact_pii=True,
    keyterms_prompt=["ICD-10", "CPT", "prior authorization"],
)

transcript = aai.Transcriber().transcribe("./encounter.wav", config)
print(transcript.text)

Structured output

Entity detection across 50+ types turns audio into mappable fields for coding and claims.

Speech Understanding

PHI redaction built in

Redact PHI across audio and transcripts in the same request.

Medical Mode

Enterprise deployment

EU data residency, self-hosted in your VPC, unlimited concurrency, forward-deployed engineers.

Universal-3.5 Pro

Common questions

: Yes — entity detection (50+ types) runs alongside transcription, giving mappable entities for EHR fields and coding.
: Yes — PHI redaction across both audio and transcripts, in the same request.
: Yes — Voice AI Cloud, EU data residency (api.eu.assemblyai.com), or self-hosted in your AWS VPC.
: Run your own audio against the benchmarks via Applied AI — we'll measure MER on your real encounters.
: Business associate under HIPAA with a standard BAA, PHI redaction, and SOC 2 Type 2.
: Pay-as-you-go per second, no minimums; Medical Mode is a $0.15/hr add-on.

Run benchmarks on your own data

Measure Medical Mode's Missed Entity Rate on your real encounters with our Applied AI team.

Talk to Applied AI