yourang Logo
Back to the platform
Documentation
Get started
  • What the platform is
  • First access
  • Platform overview
  • Quick glossary
Dashboard Assistant
  • What the dashboard assistant is
  • How to use it
  • What it can do
  • Safety and privacy
  • Plans and usage limits
Voice agent
  • GPT Live: voices, style and listening
  • Basic configuration
  • Models and voices: use cases
  • Instructions
  • Prompt tokens
  • Advanced settings
  • Evaluation criteria
  • Testing and playground
  • Updating instructions
  • External Tools
  • Built-in tools
  • MCP servers
Models offered
  • What models are and why they matter
  • Voice models
  • AI models
  • Choosing the right combination
Documents and knowledge base
  • What the knowledge base is
  • Uploading and managing documents
  • How search works during a call
  • Best practices
Calls
  • Call history
  • Transcripts and summaries
  • Audio recordings
  • Filters and search
  • Data export
Call transfers
  • When and why to transfer a call
  • Departments
  • System tools
  • AI after-hours
  • Routing rules
  • Operator app (iOS / Android)
WhatsApp
  • WhatsApp Business in yourang
  • Real-time chat
  • Automations
  • Approved templates
  • WhatsApp contacts and lists
  • WhatsApp AI agents
Meta Agent
  • The Meta agent in yourang
  • Prepare your Meta accounts
  • Connect Instagram and Facebook to yourang
  • Create the Meta agent
  • Configure the agent
  • Direct messages
  • Comments and moderation
  • Meta agent tools
  • Inbox and takeover
  • Meta in workflows
  • Plans, interactions and limits
  • Troubleshooting
Actions and campaigns
  • What actions are
  • SMS campaigns
  • Email campaigns
  • Scheduling and batch sends
Reservations
  • Calendar view
  • Availability rules
  • Confirmations and reminders
  • Changes and cancellations
Contacts
  • Customer directory
  • CSV import
  • Lead form
  • Lists and segments
  • Custom fields
Shop and catalogue
  • The shop in yourang
  • Product and service catalogue
  • Order management
  • OCR and price-list import
Integrations
  • Connect Apple Calendar
  • Connect HubSpot
  • Integrations overview
  • Calendar
  • WhatsApp
  • Meta (Instagram and Facebook)
  • SMS and email
  • Business software and PMS
  • Outbound webhooks
Workflows
  • What workflows are
  • Nodes and blocks
  • Triggers and webhooks
  • Practical examples
Call center and dialer
  • What the yourang call center does
  • Outbound campaigns
  • Human operators
  • Contact lists and live sync
  • Operator panel
  • Weekly schedule for incoming calls
  • Contact list rules
External APIs and developers
  • yourang for developers
  • API keys and authentication
  • Main endpoints
  • Incoming webhooks
Use cases
  • Hotels and accommodation
  • Restaurant
  • B&Bs and short-term rentals
  • Beauty center and spa
  • Travel agency
Pricing and plans
  • How pricing works
  • Subscription plans
  • Wallet and credits
  • Consumption and invoices
  • Changing, suspending, or cancelling the plan
Management
  • Account and organization
  • Billing and subscription
  • Team and roles
  • Security and privacy
  • Notifications
Business information
  • Business details
  • Location and address
  • Opening hours
  • AI assistant hours — moved
  • Departments and team
Resources
  • Frequently asked questions
  • Complete glossary
  • Support
Documentation›Voice agent›Models and voices: use cases

Voice agent

Models and voices: use cases

Compare model behaviour, choose a voice and fine-tune the settings.

Compare models in the compact selector under Model and voice. This guide explains use cases and limitations. After choosing a model, the catalogue shows compatible voices; search remains global. Save the combination before opening Advanced settings. Configure languages in Instructions.

The page presents models in a compact selector. Choose the model for the conversation you want to offer, then listen to its voices. These suggestions are starting points: several models may suit the same need. Try them with real customer requests.

The agent's voice and language are crucial to creating a natural, welcoming experience. yourang supports multiple voices and languages to fit your customers.

Choosing a model: use cases and limitations

GPT Live 1

Choose it if… you want conversations with spontaneous interruptions and corrections.

What you can adjust here: Listening cues, patience during pauses, tone and formality. Style and pauses guide behaviour through instructions.

Worth knowing: Listening cues and pauses are guidance, not guaranteed frequencies. Listening cues such as “mm-hm” do not confirm bookings or other actions. Interrupting speech does not automatically cancel a tool that has already started.

A scenario to try: Request a Tuesday appointment, then interrupt with “No, Thursday”. Check that it accepts the correction before confirming.

GPT Realtime

Choose it if… you want to adjust pauses, interruptions and speed.

What you can adjust here: Speaking speed, silence duration or turn endings based on sentence meaning. Tone and formality too.

Worth knowing: Continuous listening is available with GPT Live.

A scenario to try: Dictate a number with a pause halfway through. Check that it waits, then try correcting it while it answers.

Gemini Live

Good to know: With Gemini you also choose the platform. Gemini Stable is recommended for phone calls and keeps data in Europe; Gemini 3.8 is the newer model, but on the phone it can misunderstand short sentences.

Choose it if… you want to give customers time to think and explain.

What you can adjust here: Silence duration, response variation, tone and formality. A candidate to try for detailed requests and explanations.

Worth knowing: Numeric speaking speed cannot be adjusted here.

A scenario to try: Explain a problem, pause for two seconds, then add a detail. Does it let you finish? Try a follow-up question too.

Eleven v3 Conversational

Choose it if… you want to find a voice that fits your business.

What you can adjust here: Tone, formality and response variation. Listen to several voices saying the same greeting to find the right character.

Worth knowing: Not all provider expressiveness features are available here.

A scenario to try: Ask two voices to deliver the same welcome and respond to an unhappy customer. Assess clarity, warmth and fit with your business.

Nova Sonic

Choose it if… you want to welcome customers in the seven supported languages.

What you can adjust here: Style and response variation. Polyglot voices can retain their vocal identity across supported languages.

Worth knowing: Arabic and Chinese excluded; speed and pauses cannot be adjusted here.

A scenario to try: Start a request in Italian, then continue in English. Check that it retains the details and that pronunciation suits your customers.

Grok Voice Think Fast 2.0

Choose it if… you want to adjust speaking pace for numbers and instructions.

What you can adjust here: Speaking speed, speech sensitivity and silence duration, alongside common style settings.

Worth knowing: Pronunciation dictionaries are not available here.

A scenario to try: Ask for three steps and a code to be repeated. Try a slower speed and check that it remains clear and easy to interrupt.

Choosing the voice

Pick a synthesised voice that reflects your brand's tone. Available voices vary in gender, accent, and character.

  1. Listen to previews of each available voice
  2. Choose the one that best represents your business
  3. Test the voice in trial conversations
  4. Adjust the settings available for the selected model

Supported languages

yourang supports the following languages for the voice assistant:

  • Italian (IT), default
  • English (EN), for international customers
  • French (FR), for visitors from France and Switzerland
  • German (DE), for visitors from Germany and Austria
  • Spanish (ES), for visitors from Spain and Latin America

Multiple languages

Configure multiple languages supported by the model. The assistant recognises the language used by the customer; test language changes in the playground.

For each language, make sure the instructions are translated correctly and keep the same tone and content as the original.

Language detection & per-direction voices

The assistant recognizes the language the caller speaks among the ones you configured, and switches to it when the caller speaks it clearly in full sentences or asks for it. To avoid switching on a misheard word, mid-call it first asks the caller to repeat. Inbound and outbound assistants can also use different voices.

Was this page helpful?

PreviousBasic configurationNextInstructions