Voice Experience
Shape how your AI agent sounds, behaves, and responds to caller emotions.
What is Voice Experience?
Voice Experience is where you define your AI agent's identity and behavior. This includes choosing its voice (how it sounds), setting its personality (how it communicates), configuring conversation features like filler phrases, and managing how it responds when callers get frustrated.
What You Can Do Here
Choose a Voice
Select from a library of professional TTS voices. Preview them before committing.
Set Bot Identity
Give your AI a name, personality description, and behavioral guardrails.
Configure Conversation
Set up filler phrases, fallback responses, and barge-in behavior.
Manage Sentiment Response
Control how the AI reacts when callers are upset or frustrated.
Voice Selection
Your AI's voice is the first thing callers notice. BLEUM supports multiple TTS providers with a variety of voice options.
How to choose a voice
Open the Voice Selection section, pick a provider (Cartesia by default) and choose a voice from its list. With ElevenLabs you can also browse its voice library.
Type a sample sentence that matches your use case — like a greeting or a common response. Click "▶ Test Voice" to hear how it sounds.
Click "Save Voice" (or "Set as System Voice" for an ElevenLabs voice) to apply it. Every procedure uses this voice; each other language uses the voice set in its own tab.
Bot Identity
Bot Identity tells your AI who it is. This affects how it introduces itself, how it phrases responses, and what it avoids talking about.
- Company Name — Your business name, used in greetings and responses.
- Bot Name — What the AI calls itself (e.g., "Alex" or "BLEUM Assistant").
- Personality — A description of how the AI should communicate (e.g., "Friendly and professional, uses simple language").
- Guardrails — Things the AI should never do or discuss (e.g., "Never discuss competitor products").
Filler Phrases
Filler phrases keep the conversation feeling natural while the AI works out its reply. When the caller finishes speaking, the AI writes a short acknowledgement of what they said, in their language, and speaks it straight away. If the phrase is not ready within a moment, none is spoken.
You can set the filler mode for flow procedures and for agentic procedures:
| Mode | What Happens |
|---|---|
| Voice | The AI says a short filler phrase it wrote for what the caller said, then the reply. |
| None | No filler: silence until the reply. |
Sentiment Response
BLEUM can detect when a caller is upset or frustrated and automatically adjust the AI's behavior. This is one of the most powerful features for improving caller satisfaction.
How it works
During the call, BLEUM analyzes the caller's tone and word choice in real time. When frustration is detected, the AI can:
- Add empathy phrases — In a Flow Builder procedure, put a short acknowledgement such as "I understand this is frustrating" in front of the reply: one of your own empathy phrases or, if you have none, one the AI writes for the moment. In an agentic procedure the AI acknowledges the caller's feelings in its own words.
- Offer escalation — If the caller stays upset, offer to transfer them to a human agent.
Empathy phrases
You can customize the empathy phrases the AI uses. Add phrases that match your brand voice. With "Let AI learn which phrases work best" on, BLEUM records each time a phrase is used whether the caller sounds calmer on their next turn, and chooses the phrases that work more often. A phrase that has not worked yet is still chosen now and then, so it can prove itself.
Common Questions
Currently, the voice you set here applies across all procedures as your system voice. Multi-language support allows setting different voices for different languages.
BLEUM supports 7 languages with automatic language detection. Turn languages on under Language Support, then set each one's voice in its tab under System Voice.