Text-to-speech adapter contract

2 min readReferenceUpdated 2026-05-22

What you'll find here

The three records that compose a text-to-speech adapter result, validation rules, and the redaction boundary.

Text in, artifact reference out.

Text-to-speech results must not persist raw prompts, private payloads, adapter request payloads, raw generated speech, responses, credentials, tokens, or private media metadata.

Records

Record

Captures

Fields

TextToSpeechRequestMetadata

input

Redacted text summary · voice id · language · speaking rate · redacted metadata.

TextToSpeechArtifact

output

Media artifact id · MIME type · duration (ms) · byte count · redacted metadata · redaction status.

TextToSpeechResult

envelope

Result id · task id · adapter id · status (completed / partial / failed) · request metadata · generated artifact reference · errors · policy envelope id · evidence ids · receipt ids · redacted paths · creation timestamp.

Validation

Completed / partial

Require a generated artifact reference.

Failed

Require at least one error.

All

Require policy envelope, evidence, and receipt links.

What's next

ReferenceVoice postureThe decision that authorizes adapter calls.ReferenceSpeech-to-text adaptersThe input-direction counterpart.ReferenceMultimodal artifact referencesHow adapters cite generated speech without raw payloads.