Voice and Tuning

What it does

The Tuning page lets you control how your AI receptionist sounds and how it behaves during conversations. You can change the voice, adjust how fast and expressive it speaks, and fine-tune how it handles back-and-forth with callers — all without any technical knowledge. Changes take effect on the next incoming call after you save.

How to get there

In the left sidebar, click Tuning (the sliders icon). It is a top-level navigation item, not nested under Settings.

The Tuning page has two tabs: Voice Tuning and Conversation. Work through both before clicking Save Tuning Settings.

Step-by-step

Voice Tuning tab

This tab controls everything about how your receptionist sounds.

Voice Selection

A dropdown menu lists all available voices. You can filter by gender or accent using the controls above the list. Click the play button next to any voice to hear a preview before selecting it.

Take your time here. The voice is the first thing every caller hears, and it sets the tone for the entire conversation. Listen to a few options before deciding. Once you pick a voice and save, it applies to every call going forward.

Speed

A slider from 0.7 to 1.3 (default: 1.0).

This controls how fast your receptionist speaks. At 1.0 it speaks at a natural, conversational pace. Lower values slow it down; higher values speed it up.

Recommended range: 0.85–1.15. Outside that range it starts to sound noticeably robotic. If callers ever mention that your receptionist speaks too fast or too slow, this is the setting to adjust.

Expressiveness

A slider from 0 to 1 (default varies by voice).

This controls how animated and emotionally varied the voice sounds. At 0 it is flat and consistent — professional but monotone. At 1 it is highly varied, almost theatrical.

Recommended range: 0.3–0.4 for a receptionist. A small amount of expressiveness makes the voice sound warm and engaged. Too much and it sounds overdramatic for a business context.

Conversation tab

This tab controls how your receptionist handles the flow of a conversation.

Turn-Taking Sensitivity

A slider from 0.5 to 0.9 (default: 0.7).

This controls how long the receptionist waits after a caller finishes speaking before it responds. Think of it as the receptionist’s “patience” before it jumps in.

• Higher values (closer to 0.9): The receptionist waits longer, giving callers more time to finish a thought. Less likely to cut in while the caller is mid-sentence. Useful if your callers tend to pause or speak slowly.
• Lower values (closer to 0.5): The receptionist responds more quickly. Conversations feel snappier, but it may occasionally respond before a caller has fully finished speaking.

The default of 0.7 works well for most businesses. Only adjust this if you are hearing complaints — either that the receptionist interrupts callers, or that it takes too long to respond.

Routing Timeout

A slider from 5 to 60 seconds (default: 20 seconds).

This is how long the receptionist waits when it has tried to connect a caller to a team member and is waiting for that person to respond. After this time runs out, the receptionist stops waiting and either takes a message or offers another option.

If your team members consistently need more than 20 seconds to check a Slack notification and respond, increase this. If callers are waiting an uncomfortable amount of time on hold, reduce it.

Max Response Tokens

A slider from 128 to 1,024 (default: 512).

This sets the maximum length of any single response your receptionist gives. A lower number keeps responses brief; a higher number allows longer explanations.

In practice, your receptionist rarely hits this limit on normal calls — it naturally keeps responses short. This setting is a guardrail for unusual cases where it might otherwise ramble.

For most businesses, leave this at 512. If callers frequently say the receptionist gave them too much information at once, try lowering it to 256 or 384.

Interruption Threshold

A slider from 1 to 5 (default: 2).

This controls how easy it is for a caller to interrupt the receptionist while it is speaking.

• Lower values (1–2): A single word — “ok,” “yes,” “got it” — is enough to stop the receptionist mid-sentence. More responsive, but the receptionist may stop speaking when the caller is only acknowledging it rather than actually trying to take over.
• Higher values (3–5): The caller needs to say more words before the receptionist yields. The receptionist finishes more of what it is saying. Useful if callers tend to make short acknowledgment sounds that you do not want treated as interruptions.

The default of 2 means a single word on its own is ignored as an acknowledgment, but two or more words will stop the receptionist. This is the right balance for most businesses.

Saving your changes

Click Save Tuning Settings at the bottom of the page. A confirmation appears when the save is successful. Changes take effect on the next call — any call already in progress is not affected.

Tips

Listen before you finalize. Use the voice preview in the Voice Selection dropdown to audition the combination of voice plus speed before saving. There is no way to preview Expressiveness directly — make a small test call to your Azumo number after saving to hear the result.

Change one thing at a time. If you adjust Speed, Expressiveness, and Turn-Taking Sensitivity all at once and the next call sounds off, you will not know which setting caused the problem. Make one change, save, test with a real call, then adjust again.

Speed and Expressiveness interact. A faster speed with high expressiveness can sound frantic. A slower speed with low expressiveness can sound drowsy. The sweet spot for most receptionist voices is speed around 0.95–1.05 and expressiveness around 0.3–0.4.

Routing Timeout affects caller experience on hold. The receptionist tells the caller it is checking with someone and then waits silently. A 30-second wait can feel long to a caller who is not expecting it. If you raise this setting, consider updating your welcome greeting or special instructions to let callers know they may be on hold briefly.

Your voice cannot be changed mid-call. If a call is in progress when you save new voice settings, that call finishes with the old voice. The new voice starts on the very next call.

Troubleshooting

I changed the voice but it sounds the same on the next call. Make sure you clicked Save Tuning Settings before leaving the page. If the save confirmation appeared, the change was applied. If the voice still sounds the same after a confirmed save, check that the call was made after the save (not a call that was already in progress).

The voice preview button does not play audio. Make sure your browser is not blocking audio. Try clicking a different voice first, then coming back to the one you want. If it still does not play, try Chrome or Safari. If the problem persists, contact support at support@azumo.com.

My receptionist keeps cutting off callers. Lower the Turn-Taking Sensitivity slider toward 0.5 to make the receptionist wait longer before responding. If callers are being cut off even with a low sensitivity setting, also increase the Interruption Threshold to 3 or 4, which requires callers to say more before the receptionist stops mid-sentence.

My receptionist pauses too long after callers finish speaking. Increase the Turn-Taking Sensitivity slider toward 0.9. This makes the receptionist respond more quickly when it detects that the caller has stopped talking.

Callers say the voice sounds unnatural or robotic. The most common causes are Speed set too high (above 1.2) or too low (below 0.8), or Expressiveness set to 0. Try Speed at 1.0 and Expressiveness at 0.35 as a neutral starting point.

I updated the Routing Timeout but callers are still waiting a long time. Confirm the save completed successfully and that the next call came in after the save. Also check how long your team members are taking to respond to the routing notification — if they are not responding within the window, the timeout is not the issue. See Contacts to verify that the right team members are active and set up to receive routing requests.

I cannot find the Audio Timing tab. The Audio Timing tab is not available on standard accounts. It contains advanced settings that require technical knowledge to tune safely and is only shown to platform administrators.

Related

• Agent Configuration — change your receptionist’s name, welcome greeting, and special instructions
• Contacts — manage which team members the receptionist routes calls to
• Call History — review transcripts to hear whether tuning changes are improving conversations
• Business Hours — control when your receptionist takes messages vs tries to reach someone