Medical Dictation on Mac: The $99/Month Problem Nobody Talks About

The Math That Should Make Every Doctor Angry

I want to start with numbers, because the numbers are what made me write this article.

Dragon Medical One costs $99 per month per provider. If you are a solo practitioner, that is $1,188 per year. If you run a 5-physician practice, that is $5,940 per year. Over five years, a single seat costs you $5,940. A five-doctor practice will spend $29,700 on dictation software alone.

EmberType costs $49. One time. Forever. No monthly fees, no annual renewals, no per-seat licensing.

I am not pretending these two products are identical. Dragon Medical One has deep EHR integration, a medical vocabulary that has been refined over decades, and enterprise compliance features that a small Mac app cannot match. I will be honest about all of that in this article. But the gap between $49 and $5,940 is so vast that even if EmberType only covers 70% of what you need, the economics demand you at least consider whether that other 30% is actually worth $5,901.

When we set EmberType's price at $49, we made a deliberate choice. The AI model that powers transcription — OpenAI's Whisper — is open source and free. The processing happens on your Mac's Apple Silicon chip, so there are no server costs. There is no reason dictation software needs to cost $99 per month in 2026. The only reason Dragon gets away with it is that physicians have been told, for decades, that medical dictation is a specialized product requiring specialized pricing. That was true in 2005. It is not true anymore.

Why Doctors Need Dictation Differently

Before we compare tools, I want to acknowledge what makes medical dictation genuinely different from dictating an email or a blog post. Understanding this is important for evaluating whether EmberType — or any non-medical dictation tool — can serve a physician's workflow.

According to a 2024 Medscape physician burnout report, the average physician spends roughly two hours per day on clinical documentation. That is two hours not spent with patients. For many doctors, it is the single biggest contributor to burnout — more than patient volume, more than insurance paperwork, more than administrative overhead. Documentation is the silent time thief of modern medicine.

Clinical notes have demands that regular dictation does not prepare you for:

• Medical terminology is dense and precise. "Hypertrophic cardiomyopathy" is not a phrase most dictation engines train on. "Metformin 500mg twice daily" requires the engine to distinguish drug names from common words and get dosages right.
• SOAP notes follow a specific structure. Subjective, Objective, Assessment, Plan. Doctors think in this framework, and dictating in it requires the transcription to keep pace with a structured clinical thought process.
• Accuracy is non-negotiable. A transcription error in an email is embarrassing. A transcription error in a patient chart can be dangerous. "Hyper" versus "hypo" is not a minor distinction when it precedes "glycemia."
• Speed matters differently. A doctor dictating between patients has three to five minutes, not thirty. The tool needs to be available instantly, transcribe in real time, and get out of the way.

This is the context Dragon Medical has operated in for decades. And to its credit, Dragon was built for exactly this workflow. The question is whether the workflow has changed enough — and whether general-purpose AI has improved enough — that the $99/month monopoly price no longer makes sense.

The Dragon Medical Monopoly

To understand why dictation for doctors is so expensive, you need to understand the company behind it.

Dragon was created by Nuance Communications, which spent decades building the most comprehensive medical speech recognition system in the world. Their medical vocabulary is enormous. Their EHR integrations are deep. For a long time, there was genuinely no alternative — if you wanted to dictate clinical notes, you used Dragon or you typed.

In March 2022, Microsoft completed its acquisition of Nuance for $19.7 billion. Let that number sink in. Nineteen point seven billion dollars. Microsoft did not pay that for a dictation engine. They paid it for the healthcare data pipeline, the EHR integration points, and the embedded position in clinical workflows across thousands of hospitals. Dragon Medical is not software anymore. It is infrastructure that Microsoft controls.

Here is what the current Dragon landscape looks like:

• Dragon Medical One — $99/month per provider. Cloud-based. Requires internet. Designed for hospital and enterprise deployments. Deep EHR integration with Epic, Cerner, and others.
• Dragon Professional — $699 one-time. General-purpose dictation. No medical vocabulary optimization. No EHR integration. And notably, Dragon was discontinued for Mac in 2018 — it is Windows-only.
• Nuance DAX (Dragon Ambient eXperience) — AI-powered ambient clinical documentation. Listens to the doctor-patient conversation and generates notes automatically. Even more expensive than Dragon Medical One, with pricing typically bundled into enterprise contracts.

Notice the gap. If you are a doctor on a Mac, Dragon Medical One requires you to run it through Citrix or a virtual Windows environment. Dragon Professional does not have medical vocabulary. And none of these options cost less than $699. For a solo practitioner running a MacBook Pro, the options have been: pay the monopoly price, use a subpar tool, or type everything yourself.

That gap is why I built EmberType.

The Offline Privacy Angle — Why It Actually Matters in Healthcare

I talk about offline processing a lot in our marketing, and I know some people dismiss it as a feature for the paranoid. In healthcare, it is not paranoia. It is a federal compliance requirement.

HIPAA — the Health Insurance Portability and Accountability Act — establishes strict rules about how protected health information (PHI) can be transmitted, stored, and accessed. When a doctor dictates a clinical note into a cloud-based service, that audio contains PHI. The patient's name, their diagnosis, their medications, their history — all of it travels from the microphone to a remote server for processing.

Dragon Medical One handles this through enterprise compliance: Business Associate Agreements, encrypted transmission, SOC 2 certification, the full stack of healthcare security apparatus. And it works. Hospitals use Dragon Medical across thousands of providers because Nuance has spent years and millions of dollars earning those compliance certifications.

EmberType takes a fundamentally different approach. No audio ever leaves your Mac. When you dictate into EmberType, your voice is captured by your microphone, processed by Whisper AI running locally on your Apple Silicon chip, and the resulting text appears at your cursor. There is no server. There is no transmission. There is no cloud component whatsoever. I designed it this way because I believe the most secure data architecture is one where sensitive data physically cannot move.

Now, I need to be transparent about something: EmberType is not HIPAA-certified. We have not gone through the formal compliance process, obtained a SOC 2 audit, or established BAA agreements. That distinction matters for hospitals and large practices that need to demonstrate compliance to auditors and regulators. If your compliance officer asks "Is this HIPAA compliant?" the honest answer is that EmberType has not been formally certified.

But here is what I can tell you about the architecture: the primary risk vector that HIPAA addresses in cloud-based tools — the transmission and storage of PHI on third-party servers — does not exist in EmberType. There is no data to breach because there is no server holding your data. There is no transmission to intercept because there is no transmission. Your dictated clinical notes exist on your Mac and nowhere else.

For a solo practitioner dictating SOAP notes on a personal MacBook, that architecture eliminates most of the risks that HIPAA compliance is designed to mitigate. For a hospital system with 500 providers, formal compliance certification matters more. Know which category you fall into.

How Well Does Whisper AI Handle Medical Terminology?

This is the question every doctor asks, and I am going to answer it honestly rather than with marketing spin.

EmberType uses OpenAI's Whisper AI model for speech recognition. Whisper was trained on 680,000 hours of multilingual audio data scraped from the web — podcasts, lectures, interviews, YouTube videos, and a broad range of spoken content. That training data includes a significant amount of medical content: physician podcasts, medical lectures, health education videos, clinical conference recordings. As a result, Whisper handles medical terminology substantially better than you might expect from a general-purpose model.

In my testing with physician users during our beta, here is what I observed:

• Common medical terms — hypertension, diabetes mellitus, myocardial infarction, pneumothorax — transcribed correctly 90%+ of the time
• Common drug names — metformin, lisinopril, atorvastatin, amoxicillin — handled well. Newer or less common drugs had more errors
• Anatomical terms — thoracic, lumbar, bilateral, subcutaneous — consistently accurate
• Complex multi-word phrases — "left anterior descending artery," "type 2 diabetes mellitus with diabetic nephropathy" — usually correct, occasionally required minor edits
• Abbreviations and acronyms — this is where it gets shakier. "BID" might transcribe as "bid," and "PRN" might come through as "PRN" or "p.r.n." depending on how clearly you speak it

Compared to Dragon Medical One, which has been specifically trained on clinical dictation from hundreds of thousands of physicians over decades, Whisper will make more errors on specialized vocabulary. That is the honest truth. Dragon's medical vocabulary is its crown jewel, and a general-purpose model cannot match decades of domain-specific training data.

But here is the thing most people miss: most clinical documentation is not edge-case terminology. The majority of a progress note, a discharge summary, or a referral letter is plain language interspersed with common medical terms. "The patient presents with a three-day history of worsening chest pain, radiating to the left arm, associated with shortness of breath and diaphoresis." Whisper handles that sentence perfectly. And the physician's time spent correcting the occasional transcription error needs to be weighed against the $1,149 per year they are saving.

Dragon Medical One vs EmberType: The Honest Comparison

The table tells the story. Dragon wins where deep clinical infrastructure matters: EHR integration, enterprise compliance, medical vocabulary depth, and ambient documentation. EmberType wins where individual physicians care most: price, privacy, Mac support, and freedom from subscriptions.

Who EmberType Actually Works For in Healthcare

I have spent enough time with physician users to know exactly where EmberType fits — and where it does not. Here are the clinicians who benefit most from switching.

Solo Practitioners and Small Practices

If you are a solo doc or part of a two- to five-physician practice, Dragon Medical One's pricing is painful. At $99/month per seat, a three-doctor practice is spending $3,564 per year on dictation. That is a new MacBook every year. EmberType covers all three for $117 total — $49 each, once. If your workflow involves dictating notes in your own words and pasting them into your EHR, you do not need Dragon's deep integration. You need accurate transcription at a price that does not insult you.

Doctors Who Dictate Outside the EHR

Many physicians I have spoken to do not dictate directly into their EHR. They dictate into Notes, a Word document, or an email, then review, edit, and paste the finished text into Epic or Cerner. If that is your workflow, Dragon Medical's EHR integration — its strongest selling point — is irrelevant to you. You are paying $99/month for a feature you do not use. EmberType dictates directly into any app on your Mac. Speak into Notes, Mail, Google Docs, or any text field. Press the hotkey, speak, text appears at your cursor.

Therapists and Mental Health Professionals

Therapists writing session notes after hours have a particularly strong case for EmberType. Session notes often contain deeply personal patient information, dictated outside the clinic environment — sometimes at home, sometimes at a coffee shop on a laptop. The idea of that audio traveling to a cloud server should give any therapist pause. With EmberType, your session notes are processed entirely on your Mac. And at $49 versus $1,188/year, the economics are not even close.

Medical Students and Residents

A medical resident earning $60,000 per year is not going to spend $1,188 annually on dictation software. But residents are drowning in documentation — H&Ps, progress notes, discharge summaries, procedure notes. EmberType at $49 is less than a textbook. For a resident learning to dictate efficiently, it is an affordable entry point that runs entirely on the MacBook they already own.

Any Clinician Who Values Privacy

Some physicians simply do not want patient audio on a cloud server, regardless of BAA agreements and SOC 2 certifications. If you fall into that category — if the architecture of fully offline processing matters to you on principle — EmberType is the only Mac dictation tool that can make that guarantee. Read our deep dive on the voice assistant privacy crisis for more on why this matters.

Who Should Stay With Dragon Medical

I would rather lose a potential customer than mislead one. Here is who should not switch to EmberType.

• Large hospital systems — If you need organization-wide deployment with centralized administration, user management, and compliance reporting, Dragon Medical One's enterprise infrastructure is built for this. EmberType is a standalone Mac app, not an enterprise platform.
• Practices with deep EHR integration requirements — If your workflow depends on dictating directly into Epic, Cerner, Allscripts, or another EHR with Dragon's voice commands ("go to assessment," "select diagnosis"), EmberType cannot replicate that integration. Dragon's hooks into EHR systems are decades deep.
• Anyone who needs formal HIPAA certification — If your compliance officer requires every tool in your stack to carry formal HIPAA certification with BAA agreements, EmberType cannot provide that today. Dragon Medical can. The difference between "architecturally private" and "formally certified" matters in regulated environments.
• Clinicians using Nuance DAX — If you use or are evaluating ambient clinical documentation — where the AI listens to the entire patient encounter and generates notes automatically — Dragon's DAX product has no equivalent in our world. That is an entirely different category of tool.
• Windows-only environments — EmberType is Mac-only, built for Apple Silicon. If your practice runs on Windows machines, this is not for you.

The Workflow: How Doctors Actually Use EmberType

Since EmberType does not integrate with EHRs, the workflow is different from Dragon Medical. Here is what it actually looks like in practice, based on feedback from physician beta testers.

1. See the patient. Take mental notes or jot bullet points on paper during the encounter.
2. Open Notes, Word, or any text app on your Mac. EmberType works system-wide — it does not care which app has focus.
3. Press the hotkey. A small indicator appears. Start dictating your note. "The patient is a 52-year-old male presenting with a three-week history of progressive lower back pain, worse with flexion, no radiculopathy, no bowel or bladder changes."
4. Text appears at your cursor in real time. Review it, make any corrections (there will be occasional ones — no transcription is perfect).
5. Copy the finished note into your EHR. For most physicians, this step takes 10 seconds and eliminates $99/month from their expenses.

Is this workflow as seamless as Dragon Medical's direct EHR integration? No. But for the physicians who already dictate into a separate document before pasting into their EHR — and in my conversations, that is a substantial number — there is zero difference in workflow and a massive difference in cost.

For a deeper look at how speech-to-text works on Mac, including setup and configuration, see our complete guide. And if you want to compare more broadly, our best dictation app for Mac roundup covers the full landscape including 2026 AI-powered options.

---

Frequently Asked Questions