Voice typing versus outsourced transcription for radiology reports
Radiology reports have been dictated for a long time, but the way spoken words become text has changed. Some radiologists use voice typing, where speech recognition produces text on screen as they speak. Others send audio to a transcription service, in house or outsourced, and receive a typed draft back.
Both approaches can produce good reports. They differ in who does the work, when the text appears and who catches the errors. This guide sets out those differences plainly so you can decide what suits your practice. It does not recommend any particular supplier.
What is voice typing?
Voice typing is the use of speech recognition software to turn your spoken words into text on screen as you speak. You see the words appear in real time and can correct them immediately.
Voice typing can be provided by dedicated reporting software or by speech services built into web browsers and operating systems. Browser-based recognition is available in recent versions of Chrome, Edge and Safari, though in some browsers audio may be sent to the browser vendor for processing.
With voice typing, the radiologist is both author and editor. There is no one else in the loop, so the report can be finished in a single sitting.
- Text appears as you speak.
- You correct errors yourself, straight away.
- No waiting for a typist.
- Works best with clear speech and a good microphone.
What is outsourced transcription?
Outsourced transcription is a service where your recorded audio is sent to trained typists, outside your organisation, who type the report and return a draft for you to check and sign.
Transcriptionists who work with medical dictation become familiar with terminology, common phrases and individual radiologists' styles. Some services also use speech recognition first and have a person edit the result.
The radiologist still reviews and signs the final report. Transcription moves the typing and much of the first round of correction to someone else, but responsibility for the content does not move.
- You record audio and send it off.
- A typist produces the draft.
- The draft returns to you for review and signature.
- Some services combine software and human editing.
How do the two approaches compare on turnaround?
Voice typing usually has the shorter turnaround, because the text is ready as soon as you stop speaking. Transcription adds the time it takes for audio to be typed and returned, plus any time the draft waits for your review.
Transcription turnaround depends on the service, the time of day and the volume of work. It can be predictable, but it always adds at least one hand-off. For urgent studies, voice typing avoids the transcription wait.
Turnaround also depends on review habits. A voice typed report left unsigned for hours is no faster in practice than a transcribed one.
- Voice typing: text is available immediately.
- Transcription: adds a typing and return step.
- Both depend on how quickly the radiologist reviews.
Who corrects the errors in each approach?
In voice typing, the radiologist corrects recognition errors; in transcription, the typist corrects most of them before the radiologist sees the draft. This is the central trade-off between the two.
Voice typing shifts editing work to the radiologist. With clear dictation and good correction tools that work can be small, but it is still there. Transcription removes most of it, at the cost of waiting and of relying on someone else to interpret unclear audio.
In both cases, the radiologist remains the final check. Typists can mishear or guess at unclear words just as software can, so careful review is needed whichever route the text takes.
- Voice typing: the author edits their own text.
- Transcription: a typist edits before the author reviews.
- Either way, the radiologist checks before signing.
How do cost and practicalities differ?
Cost and practicalities differ mainly in how they scale. Transcription is typically charged by volume of audio or text, so cost rises with workload. Voice typing tools are often licensed per user or bundled with other software, so cost is more closely tied to the number of radiologists.
There are also practical matters to weigh. Transcription involves sending patient audio outside your organisation, which needs appropriate agreements and local approval. Voice typing needs a suitable microphone, a quiet space and a supported browser or application.
Prices and terms vary widely, so compare real quotes against your own volumes rather than relying on general rules.
- Transcription costs usually scale with volume.
- Voice typing costs usually scale with users.
- Consider where audio is sent and what agreements are needed.
- Check equipment and browser support for voice typing.
When might each approach suit a practice better?
Voice typing tends to suit practices that value fast turnaround and radiologists who are comfortable editing their own text. Transcription tends to suit radiologists who prefer to dictate without watching the screen and practices where a short delay is acceptable.
Many departments use a mix. Urgent and short reports might be voice typed, while long or complex reports are sent for transcription. Trainees sometimes prefer voice typing because seeing their words appear helps them refine their reporting style.
The right answer depends on your workload, your team and your preferences. A short trial of each, with honest feedback from the radiologists using them, is often more useful than any general comparison.
- Voice typing: speed and self-editing.
- Transcription: hands-off typing with a delay.
- A mixed approach is common and reasonable.
What questions should you ask before choosing?
Before choosing, ask how quickly reports need to be available, who will correct errors, where audio and text will be handled, and how each option fits the way your radiologists already work. Clear answers to these questions usually point towards one approach or a mix.
It also helps to involve the people who will use the system every day. Radiologists, typists and administrative staff each see different parts of the workflow, and their views often reveal practical issues that a feature list does not.
- How soon do referrers need the report?
- Who will correct recognition or typing errors?
- Where will audio and drafts be processed and stored?
- What equipment and browsers are available?
- How will the choice be reviewed after a trial period?
Where does Medicate fit?
Medicate is a voice typing tool with an added structuring step: you dictate in the browser, it arranges your words into Indication, Findings and Conclusion, and you review, correct and sign. There is no transcriptionist in the loop.
It is designed to reformat only, never to add findings, measurements or laterality you did not dictate. It shows the structured draft next to your raw transcript and highlights laterality words and numbers that do not appear in what you said. You can correct by voice commands or by typing, then download a PDF or Word file with your letterhead.
If your browser does not support dictation, such as Firefox, you can type the report and use every other feature.
Frequently asked questions
Is voice typing accurate enough for radiology reports?
Accuracy depends on the software, your microphone, your environment and how clearly you speak. Any errors must be caught at review, which applies to transcribed reports too.
Does outsourced transcription remove the need to review?
No. The radiologist still reviews and signs every report, because typists can mishear or misinterpret unclear audio.
Can I use both approaches?
Yes. Many practices voice type urgent or short reports and transcribe longer ones. Choose the mix that suits your workload.
Which browsers support voice typing in Medicate?
Recent versions of Chrome, Edge and Safari support dictation. Firefox does not, but typing works in every browser.
Medicate is a demonstration product. All data on this site is fictional.