Law firms have been dictating for decades. The old setup was a tape recorder, a secretary, and a transcription pool somewhere in the building. The new setup is increasingly a cloud voice-to-text app, your laptop, and a quiet office. Same outcome, faster turnaround, no secretary required.
There is one thing the new setup quietly added that the old one didn't have. A US technology company sitting in the middle of the data flow.
For most professional writing, that is a fine trade-off. For solicitors, barristers, paralegals, legal secretaries, and in-house counsel, it is the kind of trade-off that deserves a longer look. The material you are dictating is rarely "just" text. It is often privileged. It is often commercial-in-confidence. Sometimes it is both. And the regulatory expectations around how you handle it are not nearly as flexible as the marketing on a cloud dictation app would suggest.
This is a post about a different way of doing it. The work is the same. The speed is the same. The text appears in the same Word document. The difference is that the audio never travels anywhere it should not, because it never travels anywhere at all. If you have been searching for voice typing for solicitors UK, GDPR-compliant dictation software for law firms, or private dictation software for legal professionals, this is the proper answer.
The short version: PeekoType is a Windows voice typing application that runs entirely on your own computer. Your dictation audio never leaves your machine. No cloud account, no telemetry, no international data transfers. Just the dictation you have always done, with modern AI accuracy.
The confidentiality problem in plain terms
Cloud-based voice typing apps work by streaming your microphone audio to a server (usually in the US), running a transcription model on it there, and sending the text back. The audio is typically deleted after processing, the company will tell you. They will also tell you they are SOC 2 certified, GDPR compliant, ISO 27001 audited, and have a DPA available for enterprise customers.
All of that is probably true. None of it changes the underlying fact that the audio of you, a solicitor, dictating an attendance note about a client matter has been transmitted across an international border to a third-party processor.
Three angles on why this matters.
Legal professional privilege
Privilege protects communications between solicitor and client. It survives most things. What it does not survive cleanly is voluntary disclosure of those communications to a third party who is not subject to your duty of confidentiality. When you dictate the substance of a privileged communication into a cloud transcription service, you have introduced a new party to that communication. The transcription provider's privacy policy may say they will not share the data, but the data has still been outside the solicitor-client relationship.
It is not necessarily a privilege-defeating step in itself. The law on inadvertent disclosure is forgiving in many circumstances, and most cloud providers contractually bind themselves to confidentiality. But "we relied on a cloud transcription provider's contract" is a less comfortable position than "the audio never left our office."
SRA Principle 7 and the Code of Conduct
Solicitors regulated by the SRA are required to keep client affairs confidential, with carved-out exceptions. The Code is light on prescriptive detail about technology choices, which is appropriate, but the practical expectation is that you make defensible decisions about how confidential material is handled. "We used a US cloud service because it was convenient" tends to be a harder defence to argue than "we kept it on our own machines."
The DPA 2018 and the UK GDPR
The standard data protection points apply. You are a data controller, your client information is personal data, and processing it through a third-party cloud service means engaging a processor. That brings DPA paperwork, sub-processor lists, transfer mechanisms (the UK-US Data Bridge, in most cases), and the unspoken assumption that the framework will hold up over time. The Privacy Shield collapse in 2020 shows that frameworks do sometimes get invalidated, sometimes at very short notice.
Cloud dictation can be done compliantly. Solicitors do it every day. The question is whether you want to be the one explaining the compliance position if anything ever goes wrong, or whether you would rather not have created the compliance problem in the first place.
On-device dictation: same workflow, no third party
PeekoType is a voice typing application that runs entirely on your Windows PC. The audio of your dictation never leaves your computer. The transcription is done locally by a copy of OpenAI's Whisper model installed on your own machine. There is no server in the loop. There is no account. There is no cloud anything.
For anyone used to cloud dictation, the experience is identical to what you have been doing. Click into your Word document, or your case management system's note field, or your time recording entry, or your Outlook draft. Press F9. Speak. The text appears where your cursor is. Press F9 again to stop. The text is generated by your laptop, written into your document, and the data flow ends there.
What changes from a compliance perspective:
- No third-party processor. You are not engaging anyone to process the audio on your behalf. There is no DPA to sign, no sub-processor list to vet, no transfer mechanism to rely on, because nothing is being transferred.
- No data leaves the jurisdiction. Whatever you dictate stays where you dictated it. The audio does not cross a border. There is no server-side log of the fact that you used the software at a particular time from a particular IP.
- Works offline. Travelling to a hearing? Sitting in a client meeting in a building with no signal? Dictating in chambers? PeekoType doesn't care whether you are online. It runs the same.
- No account, no email. The application doesn't require a sign-up. It doesn't collect your firm name. It doesn't know who you are. It is just software on your machine.
For an information security officer, IT partner, or compliance lead, that combination is usually the easy answer to the conversation that starts with "can we use voice typing for client material?"
Where this actually fits into legal work
The specifics matter more than the principles, so here are the day-to-day cases where solicitors and legal professionals tend to use voice typing first.
Attendance notes
The classic use case. You have just come off a call or a meeting, you have a couple of hours of work to do before the next thing, and you need to get the attendance note written while it is still fresh. Voice typing turns a fifteen-minute typing exercise into a five-minute speaking one. Open the file note template, press F9, talk through the meeting in chronological order, press F9 again, tidy up the text, save, file.
Two small tips:
- Start with the file reference and date as the first sentences. Whisper handles spoken file references and reference numbers well, though longer alphanumeric codes are still faster to type.
- For meetings with multiple participants, name them clearly as you dictate. The note reads better with names than with "he said" and "she said".
File notes and case notes
The longer-running cousin of the attendance note. Case notes tend to be more reflective, more thoughtful, and because of that, more time-consuming to type. Voice typing lets you think out loud, which often actually improves the quality of the note. Speak the analysis as you would explain it to a colleague over coffee, and the result is usually a clearer note than the version you would have typed.
Drafting witness statements
This one is more sensitive, both in terms of content and in terms of how you handle it. Witness statements often contain very personal information from the client, and they are frequently drafted from your handwritten notes of a longer interview. Voice typing into a Word document with track changes turned on lets you draft the first version quickly, then refine in writing. The privacy of the dictation itself is non-negotiable for this kind of material, which is exactly why doing it on-device matters.
Advice letters
For partners and associates drafting client advice letters, the typical pattern is to outline your reasoning, dictate the substance, then edit in writing. Voice typing handles the substance well. It tends to be slightly worse at the precise statutory references and case citations (Whisper sometimes mishears Hansard quotations or unusual case names), so a habit of dictating substance and typing citations tends to work best.
In-house counsel and commercial-in-confidence
For in-house lawyers, the privilege questions matter slightly differently (the privilege itself is narrower in the commercial context), but the commercial confidentiality issues are stronger. Internal investigation interview notes, M&A due diligence, employment matters, regulatory correspondence: all of it sits in the category of "things you really, really do not want on a third party's server."
Paralegals and legal secretaries
Paralegals doing bundle work, summarising disclosure, drafting first-pass correspondence, and writing chronologies are some of the heaviest users of voice typing in legal teams. The sheer volume of writing involved makes the speed gains compound quickly. Legal secretaries handling overflow dictation for partners can use voice typing to type up their own notes from copy typing or recorded dictation faster, and without a cloud middleman.
A few sensible boundaries
I want to be honest about what on-device voice typing does and does not change.
It doesn't make your overall IT setup more secure. If your case management system runs in the cloud, your email is hosted by a US provider, and your document storage syncs to OneDrive, voice typing is one input method into that ecosystem. The other links in the chain are still what they are. What on-device voice typing does is make sure the dictation step itself does not add a new third party to the data flow. Your firm's existing IT decisions still set the floor.
It also does not change your professional conduct obligations. The duty of confidentiality, the duty to act in the best interests of the client, and the rules around handling privileged material. These are about how you behave, not what software you use. Voice typing is just a faster way to get words onto a page. The thinking and the judgement still belong to you.
What to think about before installing
Three practical questions:
Is your hardware up to it? Whisper runs locally, which means it uses your laptop's CPU. On a modern machine (anything bought in the last three or four years), the medium-sized model runs comfortably. On older hardware, the small model still gives respectable accuracy. There is no GPU required.
What is your firm's policy on installed software? Most firms have a software approval process for partner-level laptops and case-management workstations. PeekoType is a standard Windows installer, does not require admin rights for normal use, and does not need a server-side integration. For most firms this is a straightforward IT sign-off, but worth checking before you roll it out across a department.
Does anyone else in the team want to try it? Voice typing tends to work as a personal productivity tool first and a team tool second. If a single partner or associate wants to try it for their own files, it is a £19.99 decision they can make individually. If a department wants to standardise on it, the same software runs on every team member's machine, but each user installs their own copy.
A few notes on workflow
Voice typing changes how you write more than you might expect. Some patterns that experienced users settle into:
- Dictate, then read back. Voice typing rewards the dictate-first, edit-second pattern. You don't have to produce a polished first draft. The speed of voice typing means a rough first version and a careful read-through is faster than typing carefully from the start.
- Use voice commands for punctuation. PeekoType supports spoken "full stop", "new paragraph", "open bracket", and similar commands. Once you internalise these, your dictation comes out cleaner and you barely have to touch the keyboard.
- Keep one keyboard finger near F9. The single biggest workflow improvement is making the on-off toggle feel automatic. Within a week of using it, you will trigger it without thinking.
- For longer documents, dictate section by section. A long witness statement or skeleton argument is much easier to dictate in two-hundred-word chunks with a short pause and a stretch between them. You stay sharper, and the structure tends to be cleaner.
- Pause and think out loud. Counter-intuitively, leaving the microphone running while you think tends to produce better text than stopping every time you need to consider a point. Whisper handles "um" and "uh" cleanly and the act of thinking aloud often surfaces useful phrasing.
Three things worth knowing for your compliance review
If you are trying to get a yes from a compliance officer, an IT partner, or a Data Protection Officer, three quick facts tend to answer most of the questions before they start:
- PeekoType doesn't store audio. It transcribes in real time and discards the audio buffer the moment the text is produced. There is no audio file sitting on your hard disk afterwards.
- PeekoType doesn't store transcripts. The text goes straight into whatever application you were already in. That application's storage and retention rules apply. PeekoType itself keeps no copy.
- PeekoType doesn't connect to the internet during normal use. No licence ping, no telemetry, no analytics, no update check that runs without your knowledge. If you want to update, you download a new installer. If you do not, nothing changes.
For most legal IT teams, that combination answers the bulk of the GDPR and confidentiality questions before they are asked. Our privacy policy covers the data protection position in full, and the GDPR-by-design post goes into the technical architecture for anyone who wants to verify it before procurement.
The bottom line
Solicitors have been dictating for as long as there have been solicitors. The technology has changed, but the work has not. What has changed in the last few years is that the easiest way to dictate (a cloud voice-to-text app) quietly introduced a privacy question that the old tape-recorder-and-secretary model did not have.
On-device voice typing takes that question off the table. The audio of your dictation stays on your laptop. The text goes into your document. Nobody else is involved.
PeekoType costs £19.99 once, comes with a 14-day free trial, and doesn't require an account to install. If you would like to see what it looks like in practice, the trial doesn't ask for a card. If you have questions about how it would slot into your firm's information governance picture, please send a note to support@peekotype.com.
The dictation hour gets shorter. The compliance picture gets simpler. The client material stays where it belongs.