Creative Studio · Module
Audio & Voice
Governed text-to-speech, dubbing and consented voices for audio and video
Audio & Voice produces governed voice content: text-to-speech, dubbing, consented voice use and light audio editing. It serves standalone audio such as podcasts, audio ads and voice prompts, and tracks for Video Factory and Avatar Studio.
What it does
Audio & Voice is planned as an add-on from W3. In Year 1 there is no separate module: licensed TTS voices are used as a step inside Video Factory and Avatar Studio renders.
When it arrives, it brings voice work into the same governed pipeline as the rest of Creative Studio.
- Voice library of licensed synthetic stock voices per language, brand default voices, and an origin record for every voice: synthetic, licensed or consented person.
- TTS projects that turn a script into audio with pace, pitch, pause and emphasis controls, and regenerate a single paragraph without redoing the whole track.
- Pronunciation lexicon per tenant, brand and language for brand, product, person and place names.
- Dubbing: source audio or video goes to a transcript, then translation, then a synthetic voice, then timing alignment and review.
- Consented voices: a voice built from a consenting person's samples, only under a Rights Registry grant. The voice print is stored encrypted as an object reference, and raw biometrics are never kept in the database.
- Voice-match checks compare uploaded samples against registry voice prints and public-figure lists.
- Audio disclosure: a spoken disclosure snippet or disclosure metadata per persona policy and channel.
- Audio editing lite: edit by transcript, silence removal, noise reduction and loudness normalisation.
- Export to MP3, WAV and AAC with transcript and caption files and a provenance record.
- Linguistic review with reviewer scoring for dubbed and TTS output. AI can flag likely mistranslations but never passes a review by itself.
Later waves
W4 plans include voice prompts for the One3 Voice channel and IVR, podcast episode management and RSS hosting, multi-speaker dialogue, audio ad presets, more languages as evals pass, and audio watermark support. Real-time voice conversion, music generation and sound effects are W5 ideas that depend on licences and law being verified.
Why it matters
Voice is where synthetic media risk is highest. Keeping it inside the Generation Gate means every voice has a known origin, every consented voice has a scoped and revocable grant, and every output is disclosed and costed.
Features
What Audio & Voice includes
Voice library
Licensed stock voices per language, brand default voices and an origin record per voice.
TTS projects
Script to audio with pace, pitch, pauses and emphasis, and paragraph-level regeneration.
Pronunciation lexicon
Tenant and brand lexicon per language for brand, product, person and place names.
Dubbing
Transcript, translation, synthetic voice, timing alignment and review.
Consented voices
Voices from a consenting person's samples under a scoped grant, stored encrypted as object references.
Voice-match checks
Samples compared against registry voice prints and public-figure lists.
Audio disclosure
Spoken disclosure snippet or metadata disclosure per persona policy and channel.
Audio editing lite
Transcript cuts, silence removal, noise reduction and loudness normalisation.
Export
MP3, WAV and AAC with transcript and caption files and a provenance record.
Cost estimate and ledger
Estimate before generation and actual cost posted to the ledger.
Linguistic review
Review step with reviewer scoring for dubbed and TTS output.
Voice prompts and podcasts
IVR and voice-agent prompts to the Voice channel, plus podcast publishing (W4).
Workflow
Dub a video into another language
- 1
Select the source audio or video
- 2
Generate the transcript
- 3
Translate with the tenant glossary
- 4
Generate the synthetic voice in the target language
- 5
Align timing to the source
- 6
A linguistic reviewer scores and approves the dub
- 7
Export with disclosure and provenance; cost posts to the ledger
AI
AI skills and Digital Employees
TTS with paragraph regeneration
Generates speech and reworks individual paragraphs.
Lexicon suggestions
Suggests pronunciations for brand, product and place names.
Dubbing pipeline
Orchestrates transcript, translation, voice and alignment for review.
Voice-match and public-figure checks
Blocks samples that match someone without consent.
Linguistic review assist
Flags likely mistranslations for the reviewer and never auto-passes.
Digital Employees that use this module
They act through the same governed Actions as people; risky steps wait for approval.
Frequently asked questions
Is Audio & Voice part of Year 1?
No. It is planned from W3. In Year 1, licensed TTS voices are used inside Video Factory and Avatar Studio renders.
Can I clone someone's voice?
Only a consenting person's own voice, under a scoped and revocable Rights Registry grant, with voice-match checks against public figures. Employee consent must never be a condition of employment; verify with counsel.
Is AI audio disclosed?
Yes. A spoken disclosure snippet or disclosure metadata is applied per your persona policy and the channel's rules.
See Audio & Voice in your workflow
Tell us about your team and the outcomes you want. We will walk you through the apps, the Digital Employees and the plan for your industry.