Content address over text, voice, model, and inference arguments.
UTC instant captured immediately before the generator call that produced this receipt's immutable PCM.
Generator cost basis and consumer retained with the generated bytes.
The record accompanies the exact runtime and synthesized line.
Typed production consumer and authored reason for this adoption.
Exact generated-content lane.
Why this production needs synthesized dialogue from this generator.
Authored cost basis, including an explicit local-compute basis.
Optionallicense?: stringOptional descriptive license metadata.
Optionalsource?: stringOptional descriptive source location.
OptionaltermsCheckedAt?: stringOptional descriptive date metadata.
Exact dialogue line id.
OptionallipJoin from final-byte viseme timing to the authored speaking actor.
The supported receipt contract requests an available or not-run outcome for new receipts; older receipts may omit it. No current execution reader enforces this field or applies mouth motion, and this repository has no synthesis writer.
Local Kokoro model identity.
Immutable Hugging Face model revision.
Digest of cached little-endian Float32 PCM bytes.
Phoneme chunks on the exact model-native PCM clock.
Kokoro phoneme stream joined in synthesis order.
Adapter, native backend, model-cache, and bundled-voice byte identity.
Digest of the exact bytes consumed by synthesis.
Stable package or revision-cache-relative asset name.
Model-native PCM clock before deterministic resampling.
Exact cached mono source sample count.
Receipt schema.
Mouth targets derived from the same chunk sample clock.
Kokoro voice identity.
Host-supplied content-addressed receipt for one locally synthesized Kokoro line; the schema retains generation claims without a repository synthesis writer or receipt-byte validator.