Data Processing Addendum

Effective date: 2026-09-01.

This addendum is provided for transparency and is not legal advice.

Processing scope and purpose

Speech Relay processes audio, transcripts, account identifiers, project identifiers, and usage identifiers to provide the streaming speech-to-text API and web playground and to bill for audio processed. When session storage becomes available, audio and transcripts will also be processed to maintain account history where storage is enabled.

Duration and deletion

Session storage, the retention periods described here, and the deletion controls described here are not yet available. Today no audio or transcript is stored at all.

When session storage becomes available, playground audio and transcripts will be retained for 24 hours when storage is on by default, and API audio and transcripts will be retained for 7 days only when storage is enabled for a project. Stored sessions may then be deleted by users and may be deleted on request. At the end of the agreement, Speech Relay will delete stored content on request, subject to retained billing usage records. Data export is not currently offered.

Customer responsibilities

Customers are responsible for having a lawful basis for the audio they submit and for informing speakers or obtaining consent where required.

Sub-processors

• Microsoft Azure for hosting and processing.
• Cloudflare for network, DNS, and email routing.
• Stripe for payment processing.

The current list is published on this site. Account holders will be notified before a new sub-processor begins processing customer content. If Speech Relay becomes aware of a security incident affecting customer content or account information, affected customers will be notified without undue delay.

Security measures

• Encryption in transit.
• Credentials scoped per project and revocable.
• No audio or transcript is retained by the inference service beyond the session.

Voice data and residency

Voice and biometric data are not stored. Enrollment audio is used for its session and discarded, and no derived voiceprint is retained. Speech Relay does not offer a data residency or region guarantee.

Contact and changes

For privacy or data requests, contact privacy@speechrelay.online. For other questions, contact contact@speechrelay.online.

We may change this addendum and will update the effective date when we do.