What is medical dictation?

Medical dictation is the clinician's spoken account of a visit: the findings, the diagnosis, the treatment and the follow-up. The speech becomes text that goes into the patient's record. The clinician does the describing, and something else does the typing. That something is either a person or speech recognition software, and the choice decides who corrects the words and when.

The term covers three workflows, and they differ in where the editing happens. Our guide to medical dictation explains each one with the evidence we could check.

A transcriptionist types the report from a recording

The oldest form is the one the job title describes. The US Bureau of Labor Statistics describes medical transcriptionists as people who listen to the recorded dictation of physicians and other healthcare workers and turn it into medical reports ¹. The clinician records the notes, a person types them, and the physician reviews and signs the result.

This is often called back-end dictation, because the editing happens after the clinician has left the room.

Speech recognition writes a first draft, and a person still checks it

Many offices now let speech recognition produce the first draft. The draft is fast, but it is not finished. In a 2018 study of clinical notes dictated in 2016, the speech recognition draft had an error rate of 7.4 percent of words ². After a medical transcriptionist edited the drafts, and physicians signed the notes, the rate in the signed notes was 0.3 percent ³.

The lesson is that a clean-looking draft still needs review. That review is the step that brings the error rate down, and it takes staff time or the clinician's own time.

Front-end dictation puts the correcting on the clinician

In front-end dictation, the clinician speaks straight into the record and fixes the text as it appears. A 2013 hospital study of radiologists measured how much correction that took. Experienced radiologists corrected 6 characters per report ⁴, and new users corrected 11 ⁵. The gap suggests a learning curve, so budget time for it before you judge the tool.

An AI scribe drafts the note from the conversation, which is not dictation

An ambient AI scribe listens to the visit and writes a draft note afterward. The clinician does not speak the note on purpose, so strictly speaking this is documentation rather than dictation. Results are mixed. In a 2025 randomized trial of physicians, one scribe cut time in the note by 9.5 percent against the control group ⁶, and the other showed no significant change, at -1.7 percent ⁷. Our comparison of AI medical scribes and the question on scribes or dictation cover the choice in more depth.

The three workflows put the editing in different places

The words "dictation" and "transcription" are often used loosely. Check what a product actually does before you compare prices. Our question on dictation and transcription explains the difference in general terms.

Medical dictation holds patient information, so privacy rules apply

Patient information is protected. The HIPAA Security Rule requires an accurate and thorough risk analysis of the risks to electronic protected health information ⁸. A dictation tool that touches patient information belongs in that analysis, whether it is a transcription service, a desktop product or a scribe. Our guide to HIPAA-compliant dictation explains the contract terms to ask about.

What to do next

1. Name your workflow. Decide whether you will speak to a transcriptionist, speak into the record yourself, or use a scribe.
2. Check where the text lands. Ask whether the tool types into your record system, on the devices you actually use.
3. Test on your own notes. Dictate a set of real notes and count the errors before you commit to a product.
4. Read the product question that fits. If you are choosing software, see the question on Dragon Medical One and Dragon NaturallySpeaking. For hardware, see the question on the best microphone for medical dictation.

For the term itself, see our glossary entries on dictation and Dragon Medical One.