Voice Dictation for Doctors: Faster Notes Without the Burnout
Physicians spend an average of two hours on documentation for every hour of direct patient care. That ratio has worsened as EHRs have multiplied in complexity. The paperwork doesn't stop when the clinic closes — it follows you home. The term for this has its own clinical ring: pajama time.
Voice dictation doesn't solve everything. But it addresses the right bottleneck. Most documentation time isn't spent thinking — it's spent translating thought into typed text. Speaking is three to four times faster than typing for most people. For clinical documentation, that gap compounds across an entire career.
This guide covers how to build a dictation workflow that actually holds up in clinical practice — not just in a demo.
What dictation does (and doesn't) replace
Not every part of a clinical note benefits equally from dictation. Understanding where it helps most prevents frustration when you hit its limits.
The high-value targets are narrative sections — the parts that require original, free-form language:
- History of present illness. Typically the longest free-text section. Dictating the HPI in narrative form as you close the visit or immediately after is where most physicians see the largest time savings.
- Assessment and plan. Reasoning through differential and plan verbally comes naturally for most physicians — it mirrors how you think during the encounter.
- Physical exam findings. Observation into language is faster spoken than typed, especially for detailed musculoskeletal, neurological, or skin exams.
- Referral letters. Often templated in structure but requiring individualized content. Dictating the patient-specific portions against a standard skeleton is efficient.
- Discharge summaries. Dense, narrative-heavy, and time-consuming. Dictation significantly reduces the time to complete a summary before discharge.
- Prior authorization letters. Formulaic structure but substantive clinical content. Dictating the clinical justification is faster than typing it repeatedly across similar cases.
Dictation is less useful for structured data entry: checkboxes, drop-downs, diagnostic codes, medication lists. Those require direct interaction with your EHR interface regardless of what dictation tool you use.
Setting up a hotkey workflow
The friction of starting and stopping dictation matters more than most physicians expect. If activating the tool requires more than one keystroke, you'll stop using it during busy visits.
Resonant activates with a configurable hotkey — typically a double-tap of a modifier key or a single chord. Set it to something you can trigger without looking at the keyboard. The pattern becomes:
- Click into the target field in your EHR or notes app.
- Trigger your hotkey — one gesture, no mouse.
- Speak. Text appears as you talk.
- Trigger the hotkey again to stop.
- Correct any errors with normal keyboard editing.
The key discipline is cursor placement before you start. Dictation inserts text wherever your cursor is — in the HPI field, the plan section, a referral letter body, wherever you need it. Clicking into the right field first is the whole workflow.
How to dictate a SOAP note
The SOAP format translates well to dictation because its structure is already verbal. You can speak section headers directly and the tool will transcribe them, or you can dictate into each field separately after positioning your cursor.
A typical dictation for a straightforward primary care visit sounds like this:
Example: SOAP note dictation
“Subjective. Mr. Smith is a 58-year-old male presenting with two weeks of right knee pain following increased walking activity on a recent trip. Pain is rated six out of ten, worsens with stairs and prolonged standing, improves with rest. No locking, no giving way. He has tried ibuprofen with partial relief. Objective. Vitals stable. Right knee: mild medial joint line tenderness, no effusion, full range of motion, negative McMurray. Assessment. Medial compartment osteoarthritis, likely exacerbated by activity. Plan. Weight bearing as tolerated. Continue NSAIDs with food. Ice after activity. Referral to physical therapy. Follow up in six weeks or sooner if worsening.”
That passage — spoken naturally in under a minute — would take most physicians four to six minutes to type. And the spoken version is often better organized, because you're thinking aloud rather than transcribing after the fact.
Tips that actually make a difference
Dictation has a learning curve, but it's shallow. These adjustments accelerate it:
- Say punctuation explicitly at first. “Period,” “comma,” “new paragraph” — until it becomes automatic. Most physicians stop noticing this within a week.
- Dictate immediately after the visit. Between-patient dictation beats end-of-day catch-up. Recall is sharper, notes are more accurate, and the cognitive load doesn't stack up. Even two or three minutes of dictation while the patient leaves the room is valuable.
- Use templates as scaffolding. A pre-typed skeleton with your standard exam findings pre-populated leaves only the patient-specific narrative for dictation. Many physicians dictate only the HPI and A/P and use typed macros for the rest.
- Don't correct everything immediately. Dictate a full section, then review and correct. Stopping mid-sentence to fix a word breaks the flow and slows you down more than the error itself.
- Microphone placement matters. A MacBook's built-in mic handles dictation well in a quiet office. In a noisy environment or if you're moving between rooms, a small clip-on or desk mic improves accuracy noticeably.
Dictation and patient privacy
Most cloud-based dictation tools — including Dragon Medical One, the AI scribes, and many general-purpose tools — send your audio to remote servers for processing. That audio contains patient names, diagnoses, medications, and history. It's Protected Health Information under HIPAA, which means a Business Associate Agreement is required, and data breach risk exists even with one.
Resonant takes a different approach: everything runs on your Mac. Audio is never transmitted anywhere. Processing happens locally — on the Neural Engine in your Apple Silicon chip — and the audio is discarded immediately after transcription. No BAA is required because no third party ever receives anything.
For physicians who want complete confidence that patient information stays on their machine, local processing eliminates the risk rather than managing it.
See how Resonant compares to Dragon Medical One and cloud AI scribes →
What to expect in the first week
Most physicians describe the first few days of dictation as slightly slower than typing while the habit forms. By the end of the first week, most are at parity or faster. By the end of the first month, the time savings are significant enough that reverting feels unthinkable.
The bigger shift is cognitive. Dictating a note the way you'd explain it to a colleague — conversational, organized, spoken — is lower friction than translating thought into typed prose. Physicians who dictate report that notes end up better organized and more detailed, not just faster.
That's the actual value. Not just getting out of the office earlier, though that matters too. It's the reduction in the mental tax of documentation. Dictation lets the note follow your thinking instead of interrupting it.
Frequently asked questions
How long does it take to learn voice dictation as a physician?
Most physicians are productive within a few days. The first hour or two involves getting comfortable speaking note structure aloud rather than typing it. By the end of the first week, most doctors report dictation feeling natural. The main adjustment is learning to verbalize punctuation and section headers, which becomes automatic quickly.
Can I dictate directly into Epic, Cerner, or other EHR systems?
Yes. Resonant works with any field where you can type — including your EHR's web interface and desktop app. Position your cursor in the note field, activate Resonant with your hotkey, and speak. Text appears wherever your cursor is. No EHR-specific plugin or integration required.
Is voice dictation accurate enough for medical terminology?
Modern on-device speech recognition models handle medical terminology — drug names, procedures, anatomical terms — well out of the box. You'll occasionally need to correct a word, just as you would with any transcription tool. Most physicians find accuracy high enough that dictation is still significantly faster than typing, even accounting for corrections.
What types of clinical documents work best for dictation?
Free-text narrative sections benefit the most: HPI, assessment and plan, physical exam findings, referral letters, prior authorization letters, and discharge summaries. Structured fields (checkboxes, dropdowns) still require manual input, but the narrative bulk of most notes — which takes the longest — is where dictation saves the most time.
Does Resonant send my dictation to the cloud?
No. Resonant runs entirely on your Mac using on-device AI models. Audio is processed locally and immediately discarded after transcription. No audio, text, or patient information is transmitted to any server. No Business Associate Agreement is needed because no third party ever receives your data.
Try Resonant free
Private voice dictation for Mac and Windows. 100% on-device, no account required. Download and start speaking in under a minute.