Thoughtly provides several advanced settings to control the presence of your Voice Agent. These settings allow you to customize how your Voice Agent interacts with users, including its silence handling, speech detection, and more.

​
Sensitivity Threshold

Sensitivity threshold is a setting that determines how sensitive the Voice Agent is to any sound. This setting is crucial for ensuring that the Voice Agent can accurately detect when a user is speaking or when there is silence in the conversation. Higher the sensitivity threshold, the more confidence the Agent needs to decide something is a speech, which means the Agent will be less sensitive to background noise. The default value is 0.8, but you can adjust this based on your specific use case.

​
Small Talk

Small talk is a feature that allows the Voice Agent to engage in a small talk while some behind the scene API actions are runnning. When this feature is disabled, the Voice Agent will play music while completing API actions instead of small talk. The default is to play music, we find music to be more intuitive and engaging for users, but you can enable small talk if you prefer a more conversational approach.

​
Silence Retries

Silence retries is the number of times the Agent will try to re-engage the caller after certain period of silence. This is useful for ensuring that the conversation continues smoothly, especially in cases where the user may be slow to respond or is experiencing technical issues. The default value is 4 retries, but you can adjust this based on your specific use case.

​
Silence Timeout

Silence timeout is the amount of time the Agent will wait in silence before considering re-engaging the caller. This setting is important for ensuring that the Voice Agent does not wait indefinitely for a response from the user. The default value is 4 seconds, but you can adjust this based on your specific use case.

​
Utterance End

Utterance end is a setting that determines how much extra time the Voice Agent will wait after the user stops speaking before considering the utterance complete. This setting is crucial for ensuring that the Voice Agent accurately captures the user’s response without cutting them off prematurely. The default is Off, which is because our backend has been configured to accomodate for most speakers. However, you can enable this setting if you find that the Voice Agent is cutting off users too early or if you want to ensure that the Agent captures longer responses.