Voice macros in radiology: how to save reporting time

A radiologist working a busy abdominal ultrasound list will speak phrases like “no acute intra-abdominal pathology” and “the liver is normal in size, echogenicity, and contour” dozens of times per session. Those phrases never change. Dictating them letter by letter is not reporting. It is data entry that adds measurable minutes to every shift.

Voice macros are the fix. Speak a short trigger, get the full phrase at your cursor. The setup takes minutes per macro. The return compounds daily.

This guide covers how voice macros work in radiology dictation, which phrases to convert first, and how they interact with on-device speech processing to produce a fast, network-independent reporting workflow.

What is a voice macro in radiology dictation?

A voice macro links a short spoken trigger to a longer block of expansion text. When you speak the trigger, the dictation software replaces it with the expansion at your cursor.

The expansion can be a single finding phrase, a full normal organ description, or an entire standard report section. The trigger is typically compact and distinctive: “norm-hepatic,” “norm-chest-xr,” “add-disclaimer.” It is short enough to speak without breaking your dictation rhythm and unusual enough that you will not say it by accident inside a real report.

When the expansion fires, it types into whatever application is active: a PACS report field, a RIS text box, a browser-based reporting interface, or any other input. Because the macro operates at the cursor level, there is no app-specific configuration. The macro works wherever typing works.

If you use RadMyk for on-device radiology dictation, voice macros are included in the base product. You record the expansion text once, assign a trigger, and the macro is available immediately. No add-on, no enterprise tier.

Which phrases are worth converting to macros first?

Start with phrases you repeat verbatim more than five times per shift. Those are the candidates that return the setup time in a single session.

For most radiologists, the list quickly includes:

• Normal organ descriptions: “The liver is normal in size, echogenicity, and contour. No focal lesions identified.” “The spleen is unremarkable.” “Both kidneys are normal in size and morphology with no hydronephrosis.”
• Standard negative findings on chest X-rays: “No pneumothorax. No pleural effusion. No airspace consolidation.” “No acute cardiopulmonary process.”
• Intracranial normals: “No acute intracranial abnormality on this non-contrast study.” “The parenchymal signal pattern is normal for age.”
• Qualifier phrases used repeatedly: “previously described,” “as above,” “correlate clinically,” “recommend clinical correlation and follow-up as clinically indicated.”
• Standard technique lines: If your practice dictates the same imaging protocol description for a specific modality dozens of times per week, one macro handles all of them.
• Sign-off and liability clauses: If your institution or group uses a standard limitation statement, a macro makes it consistent every time.

The goal is not to remove your clinical voice from the report. It is to stop re-dictating the sentences that carry no individual diagnostic content. When a chest is genuinely normal, “no acute cardiopulmonary process” is a required field, not a clinical contribution. A macro fills it cleanly and frees the cognitive space for the sentences that actually need your judgment.

How long does it take to build a usable macro library?

Less than an afternoon if you approach it systematically.

Pick one report type you read at high volume. Write down the five to eight phrases you repeat in every normal or near-normal study of that type. Record those as macros in a single sitting. Dictate that report type for a week, noting which macros save time and which trigger words feel awkward to say. Adjust.

Then repeat for a second report type. After a month of this process, most radiologists have 20 to 30 macros covering 70 to 80 percent of their repetitive dictation content. The remainder, the portions that require individual clinical thought, gets dictated fresh each time.

One practical detail: choose trigger words that are not real radiology phrases. “Norm-hepatic” is safe. “Normal liver” is a phrase you might dictate literally in a report, and you do not want the macro firing mid-sentence. Compound or hyphenated triggers, or slightly abbreviated coinages, work well for this reason.

Do voice macros affect accuracy?

No. When a dictation tool processes macros locally, the recognition engine hears the trigger word, matches it to your library, and fires the expansion. There is no cloud round-trip and no additional latency beyond what local processing already involves.

The effect on effective accuracy is neutral to positive. You are dictating fewer unique words per report, which means fewer opportunities for the engine to misrecognize a long or specialty-specific term. The expansion text is pre-set and always correct.

The one accuracy consideration is choosing a trigger the engine reliably catches. If your accent or microphone characteristics make a particular trigger word hard to recognize, you will see false misses: you say the trigger and nothing fires. The fix is to choose a different trigger word, not to change how you speak. RadMyk’s guided calibration step adapts the model to your specific voice and microphone setup, which reduces this kind of miss after the initial calibration session.

RadMyk achieves 96.1% word accuracy out of the box on radiology vocabulary, measured in real conditions rather than vendor-ideal studio setups. At that baseline, trigger recognition is reliable from day one.

Can voice macros replace full report templates?

No, and they are not trying to.

A full report template is a pre-structured document for a specific type of study: a standard chest CT template, a structured abdominal ultrasound form, a brain MRI protocol with dedicated fields for each structure. If your PACS or RIS supports structured reporting and your department uses it, those templates live in the reporting system.

Voice macros live in the dictation layer. They fill individual phrases and sentences within whatever interface you are dictating into, including inside a structured template. The two tools work at different levels. A structured template organizes the report. A voice macro fills a specific field or finding within it.

If your practice uses free-text dictation (the most common approach outside of structured-reporting-mandated hospital systems), voice macros handle the repetitive content and you dictate the clinical content fresh. If your practice uses structured templates, macros speed up the individual field completion that still requires dictation even inside a structured form.

For a full picture of dictation tools and how they compare with radiology reporting platforms, the medical dictation software guide covers the landscape clearly.

What does a macro-based reporting workflow look like in practice?

A concrete example from an abdominal ultrasound session.

You open the first study. The liver looks normal. You speak “norm-hepatic.” That fires: “The liver is normal in size, echogenicity, and contour. No focal lesions identified.” Spleen is fine: “norm-spleen” fills in your standard phrase. Kidneys look normal: “norm-kidneys” completes. You add two sentences about a mild gallbladder polyp finding that requires individualized language. Gallbladder otherwise looks fine: one more macro for the standard normal comment.

The entire normal portions of the report took five spoken words. The clinical content, the part that required your eyes on the images, is the only portion you actually dictated. That is the correct division of labor between a dictation tool and a radiologist.

At 96.1% word accuracy and approximately 220 words per minute, RadMyk keeps up with full-speed dictation without making you slow down for corrections on the macro triggers.

How does on-device processing affect macro performance?

On-device dictation keeps macros working regardless of network state. If your reading room internet drops mid-session, or if you are reading in a location with unreliable connectivity, macros based on local processing continue to fire normally.

Cloud dictation tools send audio to a remote server. A dropped connection stops the session. On-device tools process locally: there is nothing external to interrupt. The macro library is on your machine. The recognition model is on your machine. Dictation continues.

For high-volume sessions where every minute matters, the absence of latency compounds in the same direction as macros: both reduce the time between your voice and the text appearing on screen. The factors that make radiology dictation faster covers how latency, accuracy, vocabulary, and macros work together to change the shape of a full day of reporting.

Getting started with voice macros in RadMyk

RadMyk includes voice macro support in the base product, for all license tiers. You record the expansion text once, name the trigger, and the macro is live immediately.

The 28-day free trial (no credit card, no setup fee) is enough time to build an initial library and measure the time savings in a real reporting session. RadMyk is free for the full duration of radiology training, so residents and fellows can build their macro habits from the start, before they qualify and begin reading at volume.

Radiologists evaluating options after the PowerScribe 360 end-of-life window will find the broader evaluation context in our radiology dictation software guide.

For pricing: RadMyk is $199 one-time for the first 100 users (rising after that), including voice macros, guided calibration, and all future updates. Full details are at radmyk.com/pricing.

Voice macros are one of the most underused features in radiology dictation. The setup cost is low. The return is consistent. After a few weeks, the phrases you used to re-dictate from memory every single shift will not even cross your mind.