πVoice Emotions
Give your voice agent emotional expression. Pick an expressive voice using the Emotionality rating, and see the emotions your agent used right inside the call transcript.
Last updated
Give your voice agent emotional expression. Pick an expressive voice using the Emotionality rating, and see the emotions your agent used right inside the call transcript.
Voice Emotions lets your AI voice agent speak with emotional tone instead of a flat, monotone delivery. Expressive voices can sound warm, excited, empathetic, or firm depending on the moment in the conversation, and you can review exactly which emotions the agent used directly in the call transcript.
Emotions are expressed by your agent's voice β they are chosen to fit the conversation as the agent speaks. This is not sentiment analysis of the customer; the tags you see in the transcript reflect how the agent delivered each line.
Not every voice can express emotion. When you pick a voice, the Select Voice window shows an Emotionality rating for each voice so you can tell at a glance how expressive it is.
The rating is shown as a row of up to five faces β the more faces that are filled in (green), the more emotionally expressive the voice. Voices that don't support emotional expression show a dash (β) in the Emotionality column.

π‘ Tip: Pick a voice with a higher Emotionality rating when your use case benefits from a warmer, more human delivery β for example surveys, customer care, or sensitive conversations.
When your agent uses an expressive voice, its spoken lines in the call transcript are annotated with emotion tags β small colored labels that show the tone the agent used for that part of the message.
Your agent can express the following emotions:
Content
Calm, warm, reassuring tone
Excited
Enthusiasm, good news, positive energy
Sad
Empathy, apologies, sensitive topics
Angry
Firm or assertive responses (used sparingly)
Scared
Concern or a sense of urgency

A few details on how the tags display:
The first time an emotion appears in a transcript it shows as a full colored tag. If the same emotion comes up again later, it appears as a small colored dot to keep the transcript easy to read β hover over the dot to see its label.
A neutral tone is the default and is not tagged, so untagged text simply means the agent spoke neutrally there.
Last updated