WebWoW_Audio#
The production area for spoken audio along the entire customer journey: phone line, website, and chat speak with one voice — from one script system, through one engine, with built-in quality analysis. In the CallFluence console, WebWoW_Audio is its own section; this chapter documents the API underneath it.
What WebWoW_Audio covers#
| Moment | Touchpoint | Channel |
|---|---|---|
| Customer calls in | call.inbound.greeting | Phone |
| Customer is on hold | call.hold | Phone |
| Call ends | call.hangup | Phone |
| Customer visits the website | web.visit | Web |
| Lead submitted, success page | web.lead.success | Web |
| Booking a meeting | booking.meeting | Phone |
| Callback scheduled | booking.callback | Phone |
| Live chat opened | chat.welcome | Chat |
Architecture#
- Script registry — every touchpoint has a German default script with placeholders (
{{firma}},{{domain}},{{telefon}}). Tenants can override any script; the origin (defaultortenant) stays visible. - Analysis — every script is checked deterministically: estimated speaking time against the touchpoint's target, customer perspective, proper umlauts, no duplicated AI disclosure.
- Render — script to audio (MP3) through the Cartesia engine: directly via
api.cartesia.aiwith your own key, otherwise automatically through the verified Together path. - Voice-to-text — audio to transcript via Cartesia Ink or Whisper. The same path the call transcription uses.
AI disclosure (EU AI Act Art. 50)#
For every AI-spoken phone touchpoint, the dialog engine deterministically prepends the disclosure sentence. It intentionally does not live in the scripts — the analysis even warns when a script would duplicate it.
Recording and transcription are customer switches#
Whether calls are recorded and transcribed is the tenant's own decision — switchable per tenant via the audio settings. Default: on.
Continue with the touchpoints in detail or go straight to the script API.