Memorat
memorat, in folklore studies: a first-hand account, told by the person who lived through it. As opposed to a fabulat, a story passed on at second hand.
The interview stays on your computer.
Transcription, speaker recognition and pseudonymisation for research recordings. Everything is computed locally, with or without a graphics card. The recording is never uploaded, so there is no processing agreement, no transfer assessment and nothing to describe in your data management plan.
Substitutions proposed: 5 You approve each one separately. Nothing is replaced automatically.
VAT invoice · 14 to 30 day payment terms · pro forma invoice on request
What it does
Every feature listed here works in the build you download. Nothing on this page is an announcement.
What it does
Local transcription
Whisper with automatic model selection and noise reduction for field recordings. Handles files from voice recorders, phones and WhatsApp.
Speaker recognition
Who speaks when, in one-to-one interviews and in focus groups. Name the speakers yourself („Interviewer”, „Participant”) and the names carry through to the export.
Recording remote interviews
Record an interview held over Teams, Zoom, Meet or Discord together with the system audio, without a second device running in the background.
Pseudonymisation with an audit trail
National ID, tax and bank numbers, phone numbers and e-mail addresses are matched by checksum rules. Names, institutions and places by a language model. You approve every change.
Consistency across a project
The same person gets the same code in every file of the study, including inflected forms. Plain find-and-replace cannot do that.
A record for the ethics committee
An exportable list of substitutions: what, into what, how many times, by which method detected and whether a human confirmed it. A document you attach to your ethics application.
Export for analysis
docx, txt and SRT with timestamps and speaker labels, plus a format for import into MAXQDA and NVivo. Original and pseudonymised versions kept separate.
Library backup
Your whole corpus, with transcripts, speakers and the substitution record, into a single file. Move it to a new computer and carry on.
The program
Captures from the running build, not visualisations. Everything in them was invented for this page.
What it looks like
Look at what is missing from that screenshot. The program found the city, the street, the first name and the hospital, but it did not detect the surname following the word “doctor”. We did not retouch the capture, because that is precisely what the review step is for: you read the text and add what the model did not see.
The arithmetic
Rates for having a single 60 minute interview transcribed by a person, taken from publicly advertised prices.
What a study costs
Move the slider and see where the difference lies. The licence is paid once and does not depend on how many hours of recording you have.
Sequence
How the work goes
Add a recording, or record the conversation
A file from a voice recorder or a phone, or a remote conversation recorded directly in the program.
The transcript is produced
Locally, on your processor or graphics card. No cap on minutes and no charge per hour of material.
Separate the speakers and name them
The program splits the recording by voice, and you decide who is who.
Review the proposed substitutions
One detection at a time: accept, reject or correct it. You can also add what the program missed.
Export the transcript and the record
The pseudonymised version for analysis and publication, the original stays with you, the substitution record goes into your study documentation.
Caveats
This section is here on purpose. A tool that promises too much costs you your credibility in front of a committee, not us.
What we do not promise
We do not claim this is the most accurate transcription available. Public benchmarks show that commercial cloud services can be more accurate. What we offer is the best accuracy available without the recording leaving your computer, and we assume you will read the transcript and correct it.
This is not anonymisation, and it is not „GDPR compliance”. The program supports a pseudonymisation workflow, which is reversible by definition. Your responsibility as data controller stays with you, and no software changes that.
Detection is fallible. Inflected forms especially: a full name in an oblique case is usually caught, a bare first name after a title often is not. That is why the program shows you every proposal for approval, never substitutes anything on its own, and lets you add what it missed.
Why local
Work that disappears
Processing research data in the cloud is not illegal. It merely requires a processing agreement, a consultation with your data protection officer, a transfer assessment and a description in your data management plan. All of that work disappears when the recording never leaves the computer. For special category data, meaning health, beliefs and sexuality, and therefore most interesting qualitative research, the requirements grow.
| Task | In the cloud | Here |
|---|---|---|
| Data processing agreement | required | not applicable |
| Transfer assessment outside the EEA | required | not applicable |
| Processor described in the DMP | required | not applicable |
| Cost per hour of recording | grows with every file | zero |
| Works without internet | no | yes |
Pricing
For comparison: having a single 60 minute interview transcribed by a person costs roughly 50 to 100 EUR. A study with twenty interviews runs to 1 000 or 2 000 EUR.
Pricing
A perpetual licence with one year of updates. Versions released during that year keep working without any time limit, including after it ends. Renewing for another year costs the same and is entirely optional.
Free
0 EURTranscription up to 30 minutes per file. Export to txt, docx and SRT. No speaker recognition and no pseudonymisation.
no time limit · the same program, without a key
Doctoral
69 EURAll features, one seat.
net · on proof of status · 12 months of updates
Researcher
199 EURAll features: speaker recognition, pseudonymisation with an audit trail, remote interview recording, library backup.
net · perpetual licence · 12 months of updates · VAT invoice
Department
749 EUREverything in Researcher across five seats, priority support and a deployment guide for your IT department.
net · 5 seats
Prices exclude VAT. Customers in the EU with a valid VAT number are invoiced under the reverse charge procedure.
Download
Windows 10 or 11, 64-bit. A graphics card speeds transcription up but is not required: every feature runs on the processor alone.
Download and run
The installer contains the complete set of models, so the program works immediately after installation, with nothing to download and no account to create. The free version is exactly the same program, just without a key. After purchase you paste the key into Settings and every feature switches on at once. The key is checked locally, with no internet connection.
On first run Windows may show a SmartScreen warning, because the installer is new and has not yet built up a reputation. Choose „More info”, then „Run anyway”.
Ordering
Ordering and invoicing
For academic institutions we issue a pro forma invoice for internal approval, followed by a VAT invoice with payment terms of 14 to 30 days. The licence key is sent as soon as you order. We do not wait for the transfer to clear.