Legal
Draft structure only. Every document below is a stub with the headings the final text needs — the wording has to come from the owner’s counsel before launch.
Terms of service
Commercial terms for the web product, including plan limits and acceptable use.
Stub · pending counselWhat TalkScribe provides: transcription, editing, AI documents, subtitles and call analytics for uploaded recordings.
Sign-up, plan tiers, fair-use limits and what happens on downgrade. Numbers to be inserted once pricing is final.
You keep all rights to recordings, transcripts and generated documents. We process them to provide the service.
Recording consent is the customer’s responsibility. No uploading of material you have no right to process.
Generated documents are drafts. Items are linked to the transcript, but the customer is responsible for reviewing them before use.
Standard clauses — final wording from counsel. Placeholder for SLA on the Business plan.
Privacy policy
What data is collected, why, for how long, and who processes it.
Stub · pending counselAccount data, uploaded recordings, transcripts, generated documents, glossaries and voice profiles.
Performance of the contract for processing recordings; consent or another legal basis for voice profiles.
Placeholder — retention periods for uploads, transcripts and documents to be confirmed by the owner.
Placeholder — cloud region for the web product, including whether processing stays in the EU.
Placeholder — list of infrastructure and model providers used by the web product.
Access, correction, export and deletion, including deletion of individual voice samples and whole profiles.
Voice data notice
The biometric-data notice shown in the app before a voice profile is created.
Stub · pending counselA voice profile contains a mathematical representation of a person’s voice, used only to recognise that person in later recordings.
Creating a profile requires recording a legal basis or consent. The field is mandatory and stored with the profile.
Voice prints are encrypted and are never included in transcripts, exports, subtitles or AI documents. Only names appear in output.
Deleting a voice profile deletes all of its voice samples. Individual samples can also be removed one by one.
Voice recognition can be disabled for the account or for a single file at any time. Speakers are still separated, just not named.
On poor recordings the system asks for confirmation instead of naming a person automatically.