Industry
Research Transcription — from recording to coding-ready dataset
Qualitative research with speaker diarization, pseudonymisation and native export for NVivo and MAXQDA.
Ethics boards expect a clear separation between audio, transcript and codebook — and a verifiable EU data residency. DeepScript covers both: servers exclusively in Germany, speaker labels can be pseudonymised on export, and transcripts feed directly into NVivo, MAXQDA and ATLAS.ti as JSON or structured DOCX. A 60-minute depth interview costs €0.27 on Premium and is coding-ready within 3-5 minutes.
Compliance
- Ethics-board ready
- Art. 28 GDPR DPA, data residency in Germany, pseudonymisation workflow and audit log of every access.
- Pseudonymisation on export
- Speaker labels can be replaced with pseudonyms (Speaker_01, Speaker_02 …) before export — clear names stay only in the master file.
- Domain-tuned speaker diarization
- Speaker diarization in both tiers, with finer separation on Premium — also suitable for focus groups with 4-6 speakers.
- Native exports for QDA software
- Structured JSON and DOCX importable into NVivo, MAXQDA and ATLAS.ti — word timestamps preserved as anchors.
Typical workflows
- Transcribing depth interviews — precise speaker diarization, custom vocabulary for theory and methodology terms.
- Analysing focus groups — 4-6 speakers separated automatically, pseudonymised export to QDA tools.
- Transcribing expert interviews — weight technical vocabulary, DOCX export for the codebook.
- Teaching research and student data collection — multi-user setup with directory permissions and two-factor auth.
Recommended setup
- Model
- Premium (€0.27/hour) — highest accuracy, DACH dialect optimization, priority queue.
- Vocabulary
- Pre-load theory names (e.g. Grounded Theory, Habitus, Pierre Bourdieu), methodology labels and field-specific proper names.
- Exports
- JSON or structured DOCX for NVivo/MAXQDA/ATLAS.ti, plus a pseudonymised TXT version for appendices.
What a workflow looks like
- 1
Sign the DPA for the research institution
Sign electronically in the Trust Center — ethics boards typically require this evidence in the application.
- 2
Create a directory per project
One directory per project with retention and access rules — clean separation of pilot, main study and teaching research.
- 3
Enable Premium + domain vocabulary
Premium for nuanced speech. Vocabulary with theory terms, methods and proper names — every discipline needs its own lists.
- 4
Export pseudonymised to your QDA tool
Pseudonymise speaker labels on export, then import JSON or DOCX into NVivo/MAXQDA. Master with clear names stays access-restricted to the project lead.
Frequently asked questions
Is the data-processing stack sufficient for the university ethics board?
Yes — Art. 28 GDPR DPA, servers in Germany, ISO 27001 and audit log are the standard package ethics boards expect. A data protection impact assessment can be derived largely from this.
How does pseudonymisation work in practice?
Speaker diarization first assigns Speaker_01, Speaker_02 etc. — on export you decide whether to include clear names or keep the generic labels. The mapping table stays in a separate master file.
Can the JSON export be imported directly into NVivo or MAXQDA?
Yes. The JSON contains speaker labels, word timestamps and confidence values — both tools accept the format via their standard importer. A structured DOCX with speaker paragraphs is available as an alternative.
Can we involve student assistants without compromising data protection?
Yes. Read/write permissions can be assigned per directory, two-factor authentication is mandatory, every access is logged. At project end a directory deletion removes all access cleanly.
Is this the right fit for your firm, hospital or agency?
Three transcriptions free, no credit card. Data stays in Germany. DPA ready to sign at /trust-center.