Skip to main content
The Voice Tab provides detailed control over how your agent sounds and handles speech. Beyond basic voice selection, you can adjust speed, pronunciation, background sounds, and advanced speech detection.
[IMAGE: Full Voice tab with all sections visible]

​
Location

Left Sidebar → Agent Settings → Voice tab

​
Voice Selection

At the top, you’ll find the same voice dropdown as the Prompt Section. Change voices here or in the top bar — they sync.

​
Speech Settings

SettingControlRangeDefault
Speech SpeedSliderSlow ↔ Fast1 (normal)
Adjust how quickly your agent speaks:
  • Slower for complex information
  • Faster for energetic conversations
  • Default (1) works for most cases

​
Pronunciation & Background

​
Pronunciation Dictionaries

Custom pronunciations for words the AI might mispronounce.
ElementDescription
Pronunciations listCurrent custom pronunciations
Add PronunciationAdd new pronunciation rule
When to use:
  • Brand names with unusual spelling
  • Technical terms
  • Names of people or places
  • Acronyms that should be spelled out
Example:
  • Word: “Atoms” → Pronunciation: “AH-toms”
  • Word: “API” → Pronunciation: “A P I”

​
Background Sound

Add ambient audio to make conversations feel more natural:
OptionDescription
NoneNo background (default)
OfficeSubtle office ambiance
CafeCoffee shop sounds
Call CenterBusy call center background

​
Advanced Voice Settings

SettingToggleDefaultDescription
Mute User Until First Bot ResponseON/OFFOFFPrevents user audio until bot speaks first
Voicemail DetectionON/OFFOFFDetect if call goes to voicemail
Personal Info Redaction (PII)ON/OFFOFFAutomatically redact sensitive information
DenoisingON/OFFOFFRemove background noise from caller audio

​
Setting Details

Mute User Until First Bot Response Useful for outbound calls where you want the agent to speak first without the caller interrupting the greeting. Voicemail Detection For outbound campaigns — detect answering machines and handle appropriately. Personal Info Redaction (PII) Automatically removes sensitive data (SSN, credit cards, etc.) from transcripts and logs. Denoising Improves speech recognition when callers have noisy backgrounds.

​
Smart Turn Detection

Intelligent handling of when the agent should start speaking.
SettingControlDescription
Smart Turn DetectionToggleEnable intelligent turn-taking
Wait TimeSliderHow long to wait before responding
When enabled, the system intelligently determines when the caller has finished speaking, reducing interruptions. Wait Time: When enabled, adjust how long the agent waits after detected silence before responding.

​
Interruption Backoff Timer

Controls how the agent handles when users interrupt.
ControlDescription
SliderTime before agent resumes after interruption
Shorter = agent resumes quickly after interruption Longer = more patient, waits for caller to finish

​
Voice Detection Tuning

Fine-tune voice activity detection for challenging audio conditions.
SettingControlRangeDescription
ConfidenceSliderLow - HighHow confident system must be that speech is occurring
Min VolumeSliderLow - HighMinimum volume to count as speech
Trigger TimeSliderSecondsHow long speech must occur to trigger detection
Release TimeSliderSecondsHow long silence before speech is considered ended
When to adjust:
  • Callers with very quiet voices → lower Min Volume
  • Noisy environments → higher Confidence
  • Callers who pause often → longer Release Time

​
Saving Changes

After making changes, you’ll see: 
"You have unsaved changes"
[Discard] [Save]
Click Save to apply your changes.

​
What’s Next

Model Tab

Configure LLM and language settings

Testing Your Agent

Test voice settings