Transcription for Law Firms – § 43a BRAO and GDPR Art. 9 compliant
Word-accurate transcription of client interviews and witness statements, without breaking attorney-client privilege.
Data stays in Germany · ISO 27001 · DPA ready to sign
Glad the appointment worked out for you.
Of course. Shall we get started right away?
Yes – I'll record the conversation, is that okay?
For law firms, confidentiality is not a marketing claim but a professional obligation under § 43a BRAO. DeepScript processes all audio exclusively on its own servers in German data centres – no US subprocessor, no use for model training, no third-party sharing. Client interviews, witness statements and court recordings can be turned into citation-ready transcripts with speaker diarization and word-level timestamps. The Art. 28 GDPR data processing agreement is ready to sign; an hour of Standard transcription costs €0.18, Premium for nuanced legal terminology €0.27.
Compliance
The legal question first – then the workflow
What data-protection officers, internal audit and regulators look at first.
§ 43a BRAO confidentiality
Processing as auxiliary personnel under § 203(3) StGB. Staff are explicitly bound to attorney-client privilege.
GDPR Art. 9 – special categories
Health, criminal-record and social data processed with elevated technical safeguards (encryption at rest and in transit, audit log).
DPA per GDPR Art. 28
Pre-drafted DPA available in the Trust Center. Sub-processors (Hetzner, Stripe) are documented.
Servers in Germany, ISO 27001
Own infrastructure in German data centres, no Schrems II exposure. The full pipeline is operated under ISO 27001 certification.
Typical workflows
What organisations like yours use DeepScript for
Word-accurate transcription of client interviews and witness statements, without breaking attorney-client privilege.
- Transcribing client interviews – speaker diarization for attorney/client, word-timestamped quotes for the file.
- Documenting witness statements – precise verbatim capture, custom vocabulary for names and case references.
- Recording court hearings – multiple speakers (presiding judge, defence, prosecution) separated automatically.
- Dictating legal research notes – preparing briefs and memos by voice instead of typing.
Speaker timeline
Glad the appointment worked out for you.
Of course. Shall we get started right away?
Yes – I'll record the conversation, is that okay?
Recommended setup
How to set it up
Highest accuracy, DACH dialect optimisation, priority queue and finer speaker separation – the right choice for terminology in regulated settings.
One click, four formats
Model
Premium · €0.27/hour
Custom Vocabulary
Add case file numbers, statute references, client names, Latin legal terms and court locations to Custom Vocabulary.
Exports
DOCX and TXT for the case file, JSON for full-text search inside the firm's case-management system.
How to get started
From DPA to finished transcript
- 1
Sign the DPA
Sign the data processing agreement electronically in the Trust Center – takes under two minutes and is required before the first production upload.
- 2
Pick Premium + Custom Vocabulary
Premium model for legal terminology, plus a per-matter vocabulary listing case file numbers, parties and relevant statutes.
- 3
Review and export to the case file
Listen through in the editor, verify proper names, then export DOCX to the case file and JSON to document management.
FAQ
Frequently asked questions
Does using DeepScript violate attorney-client privilege under § 43a BRAO?+
No. We act as auxiliary personnel under § 203(3) StGB, with all staff bound to attorney-client privilege. The Art. 28 GDPR DPA contractually establishes processor instructions, and data remains on our own servers in Germany.
How long are recordings and transcripts of legal matters stored?+
By default 30 days, then deleted automatically including the audio. With the Pro subscription (€22/month) retention is configurable per directory – useful when case-file retention duties require longer storage.
Are audio files used to train AI models?+
Never. Client data does not feed into model training, neither ours nor a subprocessor's. This is explicitly excluded in the DPA.
Can transcripts be used as evidentiary verbatim records?+
Word timestamps and per-token confidence values are included in the JSON export, allowing comparison against the original audio at any time. For a formal verbatim record under ZPO we still recommend final attorney review – the transcript is a working basis, not an automatic court record.
How long does transcription of a 90-minute hearing take?+
On Premium typically 4-7 minutes. Cost is around €0.41 per 90-minute hearing.
Safe enough for regulated environments?
Three transcriptions free, no credit card. Data stays in Germany, and the Art. 28 GDPR DPA is ready to sign.