Voice Tab
The Voice tab is where you choose how your agent sounds. Bolti supports multiple TTS providers and ships with a curated set of preview-able voice cards.
Voice cards
The main view is a grid of voice cards. Each card shows:
- Name (e.g. Aria, Marcus, Anushka)
- Gender badge (color-coded)
- Language
- Characteristics (e.g. warm, professional, energetic)
- Provider (shown on hover)
- A ▶ play button that streams a 3-second preview from the actual TTS provider, in the voice's native language
Click a card to select it. The selected card gets a primary border and a check badge in the corner.
Filtering voices
Above the grid:
- Search — type a name fragment or voice ID
- Gender filter — All / Male / Female
- Language filter — All / English / Hindi (extend as needed)
- Custom toggle — opens advanced provider/voice ID editing (see below)
Previewing
Clicking the play button on any card streams an actual TTS sample so you can hear the voice before committing. Sample text is automatically chosen to match the voice's language (e.g. "Hello! I am Bolti" for English, the Hindi equivalent for Hindi voices).
Click again to stop. Switching voices stops the previous preview automatically.
Supported providers
| Provider | Voices | Notes |
|---|---|---|
| Cartesia | Blake, Daniela, Jacqueline, Robyn (and more) | Sonic-3 model, low-latency, multilingual |
| ElevenLabs | Alice, Chris, Eric, Jessica, Adam, Sarah, and many more | Eleven Turbo v2.5, ultra-realistic |
| SarvamAI | Anushka and other Indian-language voices | Best-in-class for Hindi and Indic languages |
| SmallestAI | Irisha and others | Lightweight, fast |
| Inworld | Ashley, Marcus, Bella | Conversational character voices |
The voice cards in the grid are a curated subset across these providers. To use a voice that isn't in the grid (e.g. a custom ElevenLabs cloned voice), use Custom mode below.
Custom mode (advanced)
Click the Custom toggle in the filter bar to open the advanced panel. It exposes:
- Voice Provider dropdown — switch the underlying provider directly.
- Manual Voice ID field — paste any voice ID supported by the selected provider (e.g. an ElevenLabs voice ID for a voice you cloned in your own ElevenLabs account).
Enable the Manual Voice ID checkbox to type a value. While disabled, the voice ID is locked to whatever's currently selected in the card grid.
For ElevenLabs, the format is <model>:<voice_id> — e.g. eleven_turbo_v2_5:Xb7hH8MSUJpSbSDYk0k2.
For Cartesia, the format is <model>:<voice_id> — e.g. sonic-3:a167e0f3-df7e-4d52-a9c3-f949145efdab.
Speed, pitch, volume
Below the voice grid:
| Field | Range / type | Default | When to change |
|---|---|---|---|
| Speed | 0.5x to 2.0x slider | 1.0x | Slow down for clarity in support; speed up for high-volume sales calls |
| Pitch | Number, provider-specific scale | Provider default | Almost never; only when your chosen voice sounds slightly off |
| Volume | Number, provider-specific scale | Provider default | Same — leave at default unless you've measured a problem |
Speed is the only one of these you'll routinely touch. Pitch and volume defaults are tuned per provider; tweak them only after you've heard a clear issue in production calls.
Picking a voice — practical advice
- Match persona to voice characteristics. A Sales persona usually wants energy (
energetic,warm). A Helpdesk persona wants calm (professional,calm). A Healthcare persona wants gentle (warm,gentle). - Match language to provider strength. SarvamAI for Hindi, Cartesia or ElevenLabs for English, ElevenLabs for European languages, Inworld for character-driven voices.
- Listen before you ship. The play button is your friend. Always preview at least 2–3 candidates before committing.
- Test in real conditions. Voices that sound great in the dashboard preview can sound different over a phone codec. Make a real call (Make Outbound Call → dial yourself) before going live.
Cloning your own voice
To use a voice you've cloned (e.g. on ElevenLabs) instead of the curated cards, you have two options:
- Custom mode — paste the voice ID directly (above).
- Import voices — bulk-register your voices with Bolti so they appear in the card grid. See Customizations → Voice Cloning & Import.