Nova
OpenAIEnergetic, lively female voice.
A warm, approachable persona with the fresh first impression of a 20s–30s agent. Easy, friendly, and unintimidating.
Customer
Can I talk to someone now?
BringTalk Agent
Friendly Rookie Agent handles the tone, context, and next action.
Intent detected
Done
Persona tone applied
Applied
Audition 4 matched voices for this persona
Review response start, model cost, and persona copy together to check whether the call can reassure the customer and create the next action.
VOICE QA QUEUE
Listen to each candidate and narrow tone, pace, and handoff style for the scenario.
Energetic, lively female voice.
Calm and clear young adult female. Narration.
Strong, confident female. Narration.
Bright, friendly female voice. Casual support.
Customer
I want help, but I am not ready to decide right now.
BringTalk
Friendly Rookie Agent lowers friction and leaves a next-contact or booking option.
Reduce friction
Applied
Suggest next action
Drafted
REFERENCE COST
This estimates only speech synthesis model cost from call volume and talk time. Implementation, integration, and QA operations are scoped separately.
Operating inputs
Adjust call volume and agent speech share to estimate model cost only.
agent minutes/call = average call length × AI talk ratio
chars/call = agent minutes × 250
annual chars = chars/call × annual calls
annual USD = annual chars ÷ 1000 × $/1k
annual KRW = USD × 1370Lowest annual TTS cost at this volume
₩770,625
≈ $563 /yr
Lowest model
Ash
OpenAI
37,500,000 chars
Annual characters
150,000 min
Annual agent speech
Synthesize voice