The workflow, step by step
Every voice to text approach for care notes passes through the same six stages. Where tools differ is in how many of them they do, and how much is left to you.
The visit or conversation
A home visit, an assessment, a review, a phone call. The information exists in your head and in what the person told you.
You capture it
Either by recording the conversation with the person's agreement, or by talking through what happened afterwards in your own words.
Speech becomes text
Speech recognition turns the audio into a transcript. On its own this is dictation: your words, in order, as typed text.
Text becomes a structured draft
The transcript is organised into the sections a care record needs. This is the step that separates documentation from dictation.
You review and edit
You read the draft against what you know, correct it, add context, and rewrite the analysis in your own professional voice.
The approved record
The note you signed off is copied into your organisation's system of record. That is the record; everything before it was a draft.
Steps 1, 2, 5 and 6 are always yours: the professional judgement in the record, and accountability for it, stay with the practitioner at every stage. A phone's built-in dictation does step 3. A transcription service does step 3 and sometimes a generic summary. Structured documentation tools do step 4, and that is the one that saves the time.
Why dictation is not documentation
Dictation is old technology and it works. Speak, get text. The trouble is what you get: a stream of words in the order you thought of them, which is rarely the order a record needs. You then spend the time you saved on typing moving sentences around, splitting observation from opinion, and adding the headings.
Dictated
So I went to see Mrs E this morning about half ten, her son was there, she seemed quite low actually, said she hasn't been sleeping, the place was fine, tidy, she's eating, son said he's worried about the stairs, she said she's fine on the stairs, I think we need to get OT out, I said I'd ring them, oh and she wants to know about the day centre.
Structured
Contact: home visit, 10:30, announced. Present: Mrs E, her son. What Mrs E said: not sleeping well; feels "fine" on the stairs; would like information about the day centre. What her son said: worried about the stairs. Observed: home tidy; Mrs E low in mood, quiet, eating normally by her account. Analysis: differing views on stair safety; in my view an OT assessment would settle this and Mrs E has agreed. Actions: OT referral (me, by Friday); day centre information to Mrs E (me, by Friday).
Same visit, same words. The structured version is the one that can go on file. The gap between the two is what the structuring step closes, and it is the reason "voice to text" for care notes has to mean more than speech recognition.
Record live or dictate afterwards?
Both work, and most practitioners use each in different situations.
Recording the conversation
Best for assessments, reviews and multi-agency meetings where the detail of what each person said matters. Needs the person's agreement, and speaker separation so contributions are attributed correctly.
Dictating afterwards
Best for routine visits and calls. Two or three minutes in the car, in your own words, while it is fresh. No recording of the person, and you naturally include your own observations and thinking.
Whichever you use, the same-day habit matters more than the method. A spoken account captured within minutes of the visit is more accurate than a typed one produced hours later, and considerably more accurate than one written the next day. See how to write social work case notes for what the draft should contain.
What affects the quality of a voice draft
- Speech recognition built for conversation, not commands. Care conversations include accents, dialects, overlapping speech and medical and social care vocabulary.
- Speaker separation. In a meeting, a draft that cannot tell who said what will attribute a family member's concern to the person, or a consultant's opinion to a support worker.
- Language coverage. Conversations in UK social care happen in many languages; the tool needs to transcribe them and, where useful, work across them.
- A structure that matches the document. A case note, an assessment and a review need different sections. A tool that only knows "meeting summary" produces the wrong shape.
- A review step that is quick. Every draft needs checking. If checking is slower than typing, the tool has failed.
- Data protection. Audio and transcripts about identifiable people are sensitive. Where they go, how long they are kept, and who can see them are questions to answer before using any tool.
Where ScribisH fits
ScribisH helps health and social care professionals turn spoken or captured information into structured professional documentation, ready for practitioner review.
ScribisH covers steps 3 and 4 of the workflow above and makes step 5 quick. It uses AI to draft your notes from what you say. You can record a conversation with consent, with speaker separation and transcription in 45 or more languages, or use Dictate My Visit to talk through a visit afterwards. Either way, the draft comes back in the structure of the document you are writing, for you to review, edit by voice or keyboard, and copy across.
Based on feedback from ScribisH's first 30 practitioners, users report up to 98% accuracy and up to 80% less time spent on admin.
Frequently asked questions
You speak, either recording a conversation with consent or dictating a summary afterwards. Speech recognition turns the audio into text. A structuring step then organises that text into the sections of a care note. The practitioner reviews and edits the draft and approves the final record.
Dictation gives you your words back as typed text, in the order you said them. Structured documentation organises them into a professional record with the right sections: what happened, what the person said, what you observed, your analysis, risk, and actions. ScribisH does both steps.
No. Many practitioners prefer to talk through the visit afterwards, in the car or back at the office, and let ScribisH draft from that. Recording a conversation live is an option where it is appropriate and the person has agreed.
ScribisH uses speech recognition built for real conversations, with speaker separation for multi-person meetings and transcription in 45 or more languages. Like any speech tool it works best with a reasonable microphone and some quiet, and the practitioner always reviews the draft before it is used.
Related resources
AI note taker for social workers: what it does and how it fits your workflowHow to write social work case notes: structure, examples and common mistakesCare Act assessment template: sections, prompts and a worked exampleSocial work documentation software: what to look for and how ScribisH comparesHow ScribisH protects your dataScribisH pricingSay it once. Review it. Done.
Talk through your next visit and see the structured draft ScribisH gives you back.
Try ScribisHFree plan includes 10 notes a month. No card needed.