Configuring voice settings

📘
The Twilio integration is currently in beta. If you are participating in the beta program, we'd love to hear your feedback! Please email your thoughts and experiences to [email protected].
If you're interested in joining the beta to try out this exciting new capability, please visit our signup form here to request access. We'll notify you as soon as a spot becomes available.
During the beta period, please keep in mind that:

Some features may still be under active development

Documentation may be incomplete or subject to change

We greatly appreciate your willingness to be an early adopter and help shape the future of voice AI!

Overview

The Voice settings allows you to fine-tune the behaviour of your voice agents during phone calls. By adjusting parameters like silence timeouts, audio cues, and ASR/TTS settings, you can optimize the conversation flow and enhance the caller experience. This guide will walk you through the available options and how to configure them for your voice agent.

Getting Started

To access the Telephony settings for your agent:

Navigate to your voice agent project.
Navigate to the Settings > General tab.
Scroll down to the Telephony section.

You'll see several settings that control different aspects of the voice interaction.

Key Concepts

Silence Timeout: The amount of time to wait for the caller to start speaking before considering it a "silence" and moving on.
Utterance End Timeout: The duration of silence required after the caller starts speaking to determine they have finished their utterance.
Punctuation Timeout: The pause duration after a caller's sentence ends with a punctuation mark (period, question mark, etc.) before concluding their turn.
No Punctuation Timeout: The pause duration to wait after a caller's sentence without ending punctuation before wrapping up their turn.
Interruption Threshold: The number of words (or characters for eastern languages) detected before an interruption is considered and the agent will stop talking.
Audio Cue: A short sound effect played to the caller to signal that the agent has received their input and is processing it.

How To

Adjust Silence Timeouts

Locate the (ASR) Silence Wait setting.
Click the drop-down menu to select a value between 0ms and 5000ms.
Adjust the (ASR) Utterance End, (ASR) Punctuation Wait, and (ASR) No Punctuation Wait settings in a similar manner.
Test your agent and listen to the conversation flow to find the right balance of speed, accuracy and allowing the caller to finish speaking.

Set an Audio Cue

Navigate to the Audio Cue setting.
Click the drop-down menu and select one of the preset sound effects. Hear a preview of the cue by selecting the play button in the dropdown.
Test the audio cue by calling your agent and hearing it play after you finish speaking.

Select TTS Voice

Review the list of available TTS voices in the drop-down menu under Global Logic (top of General settings tab).
Select the voice that best matches your agent's persona and target audience.
Test the voice by calling your agent and listening to its responses.
Keep in mind you can use SSML tags to further customize the TTS output for specific prompts.

Best Practices & Tips

Start with longer timeout values (3-5 seconds) and gradually reduce them while monitoring the impact on call quality. Timeouts that are too short can lead to the agent interrupting the caller.
Choose an audio cue that is short, pleasant, and matches your brand persona. Avoid sounds that are startling, annoying or easy to confuse with common phone tones.
When supporting multiple languages, use clear prompts and confirmation flows to detect the caller's preferred language and switch the ASR dynamically (see Voice Custom Actions guide).
Explore the variety of TTS voices to find one that callers find natural and engaging to interact with. Don't be afraid to experiment with different options.

Troubleshooting

Agent Is Interrupting Caller

Increase the (ASR) Silence Wait and (ASR) Utterance End timeout values to give callers more time to start and complete their statements.
Make sure your agent prompts are worded clearly to elicit succinct responses from callers. Open-ended or ambiguous prompts may encourage rambling.

Caller Is Confused by Audio Cue

Try a different preset sound.
Consider disabling the audio cue if callers find it disruptive or unhelpful.