Industry
Transcription for Journalism — source protection, speed, 99+ languages
Turn research material into quote-ready text in minutes, without routing sources through US servers.
Source protection under § 53 StPO only works if the tooling chain stays under your control too. DeepScript hosts all recordings on its own servers in Germany, no US subprocessors — no tracking, no training use. Word timestamps make every quote traceable to 100-millisecond precision; 99+ languages cover international reporting. A 30-minute interview is transcribed on Premium for €0.14 in under three minutes.
Compliance
- Source protection § 53 StPO
- No third-party sharing, no US cloud, full audit log of data access. Account deletion removes all recordings completely.
- Servers in Germany
- Own infrastructure in German data centres, ISO 27001. Schrems II compliant with no third-country transfer.
- No tracking, no training use
- No analytics on research data, no use for model training. Default 30-day retention, then automatic deletion.
Typical workflows
- Transcribing sound-bite interviews — word timestamps for quote-ready segments, diarization between reporter and source.
- Documenting background conversations — encrypted upload, automatic deletion after the story is filed.
- Logging press conferences — fast full-text search across the finished transcript for deadline work.
- Multilingual foreign reporting — 99+ languages, optional automatic language detection for mixed recordings.
Recommended setup
- Model
- Premium (€0.27/hour) — highest accuracy, DACH dialect optimization, priority queue.
- Vocabulary
- Pre-load source names, place names, topic-specific jargon and acronyms of agencies or organisations into the vocabulary.
- Exports
- TXT for the article draft, DOCX for the desk, JSON for custom research tooling.
What a workflow looks like
- 1
Upload the recording encrypted
Drag-and-drop or API. Transfer encrypted with TLS, storage encrypted at rest — even research interns can't access others' recordings.
- 2
Pick a language or use auto-detect
Set the source language for international reporting, or auto-detect for mixed recordings. Premium model for dialects and accents.
- 3
Mark quotes in the editor and export
Jump back into audio at relevant moments via word timestamps in the editor, verify quotes, then pull TXT/DOCX into the article.
Frequently asked questions
Can authorities compel DeepScript to disclose source material?
Data is held exclusively in Germany and therefore subject to German law. § 53 StPO and German press-law privileges apply as with any German processor. We publish transparency reports on government requests.
How fast is a 60-minute recording transcribed?
On Premium typically 3-5 minutes, on Standard slightly faster. For deadline work, Premium uses the priority queue.
Are 99+ languages actually production-ready for foreign reporting?
Yes — for the major languages (EN, DE, FR, ES, IT, PT, AR, RU, ZH, JA) the Premium model consistently delivers above 95 % accuracy. Accuracy drops for niche languages or heavy dialects — custom vocabulary with place and proper names helps a lot.
Are research recordings used for training or analytics?
No. No training use, no tracking, no analytics cookies on research data. Contractually excluded in the DPA.
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.