AI agents
Choose how your agent sounds
The Voice and music tab sets the agent's voice and how it speaks, an optional background sound so it sounds like it's somewhere real, and the music callers hear while they wait.
How to choose a voice
- Open AI agents, pick the agent, and open the Voice and music tab.
- Choose a voice from the list at the top. Australian voices are listed first. The first choice is the standard voice, which is a good place to start.
- Press ▶ Hear greeting to hear the agent say its greeting in that voice.
- Try a few voices, settings and background sounds, pressing ▶ Hear greeting each time to compare.
- Press Save when you're happy. Nothing changes on calls until you save.
Next to the voice, Voice model trades speed for naturalness:
- Fastest
- The quickest replies. Great with the standard voices; a few library voices can stretch or warble words on it.
- Balanced
- About a fifth of a second slower to start, and steadier. Best for library voices.
- Best
- The most natural sound, but noticeably slower to start replying.
If you're not sure, leave it as it is.
Delivery, stability and speed

- Delivery
- Natural: talks like a person adds the odd "right" or "let me check that", a pause where a person would breathe, and a pause before reading a number back. Crisp: plain and even says the same things plainly, which suits an announcement or a strict script.
- Stability
- How much the voice may vary from one sentence to the next, from 0.2 (varied) to 0.9 (steady). Lower values add emotion and emphasis but can stretch words or wander in tone. Higher values sound the same every time, which can start to sound read out. 0.4 to 0.5 suits most phone conversations. Leave it empty for the standard setting.
- Speed
- The pace, from 0.8 to 1.15, where 1 is the voice's own pace. 0.95 to 1 sounds unhurried. Go slower only if callers need to write things down, and avoid going under 0.9 or over 1.1, which sounds stretched or rushed. Leave it empty for the standard setting.
How to add a background sound
A background sound plays quietly under the agent for the whole call, so it sounds like it's in a real place rather than a silent room. Only the caller hears it; it doesn't affect how well the agent hears the caller.
- On the Voice and music tab, choose a Background sound:
- Café
- Restaurant
- Busy office
- Call centre
- Reception: a quiet waiting area, with distant voices and footsteps
- Choose the Background level: Low, Medium or High.
- Press ▶ Hear greeting to hear the voice with the background under it.
- Press Save.
The background stops while the caller is on hold or a transfer is ringing. If the call is handed to another AI agent, that agent's own background sound (or none) plays instead. Choose None to turn it off.
Language
Your agents can speak and understand Australian English, Chinese, Japanese and other languages. The language is set for your whole business: to choose or change it, contact Binary Elements.
Hold music
Under Hold music, choose what callers hear while they're on hold with this agent, for example while a screened transfer asks the person whether they'll take the call. Press ▶ Play to listen to a track. Leave it on the first choice to use your business's usual hold music. A screened transfer can also choose its own music (see Screening).
You add and manage your own tracks on the Music on hold tab. See Music on hold.
