The Voice AI Text-to-Speech (TTS) configuration lets you control how your AI voice interacts with customers by defining voice selection, speech delivery, and text processing rules. This configuration is used in Voice AI Agents (voicebots) and IVR flows to ensure that responses sound natural, clear, and consistent.

​

This article is intended for Contact Center Administrators and Voice Solution Designers who manage voice experiences in the Voice Care.

​

For more information on permissions required to view TTS Configuration page and accessing TTS Configuration page, refer to the Getting Started with TTS and STT article.

Accessing TTS

1. Navigate to AI+ Studio from the Sprinklr launchpad.

2. Click the AI Use Cases card.

   ​

3. Navigate to Sprinklr Service tab.

4. Under Voice Workflow, select TTS. The Text-to-Speech Configurations page appears.

   ​

   ​

Configuring TTS

In the Text-to-Speech Configuration page, click + Configurations to configure the following sections:

​

The Text-to-Speech configuration page appears.

​

Basic Details

• Name (required) - Enter a unique name for the TTS configuration.
• Description - Provide a brief description to help identify the configuration.

Voice & Model

This section defines the voice and underlying AI model used for speech synthesis.

• Voice (required) - Select a voice from the integrated provider library.
  Click Explore Voices to preview available voices before selection.

• Model (implicit with voice/provider)
  The selected voice is tied to a specific provider and model. Different models vary in:

  • Voice quality
  • Supported languages
  • Latency (response time)

Tip: Choose a voice and model combination that balances natural tone and performance based on your use case.

Speech Settings

Use this section to control how the voice delivers speech.

• Language - Sets the base language for the deployment. Must match the primary language of the text content being converted.
• Text Normalization - Enable Text Normalization to standardize how common formats are read aloud.

• Auto: Model auto-detects and normalises text. Recommended for most deployments.

• Off: No automatic normalisation applied. Use only when Pronunciation Controls handle everything manually.

• On: Automatic normalisation applied.

• Stability - Controls how consistent and predictable the voice performance is. Higher values produce more stable, uniform delivery; lower values allow more expressive variation.
• Style - Amplifies the speaking style of the selected voice. Higher values exaggerate the voice's natural style; lower values keep it neutral.
• Speed/Speech rate/Style instructions/Instructions - Controls how fast the voice speaks. Values below 1.0 are slower; above 1.0 are faster.
• Similarity Boost - How closely the output should match the original voice's character. Higher values keep the voice closer to its trained identity. Example: 0.75 = close match · 0.9 = very close to original voice
• Prompt - Instructions you give to the AI model to guide how it narrates speech. Use this to provide context about your domain, type of roles. Example, "Behave in a conversational manner and take long pauses
  after commas.
• Pitch - Controls how high or low the voice sounds. Lower values create a deeper voice; higher values create a higher-pitched voice.
  Example -10 = deeper · 0 = default · +10 = higher
• Volume - Controls output loudness. Higher values increase volume; lower values reduce it.
  Example -5 = softer · 0 = default · +5 = louder
• Language Accent/Dialect - Selects the regional accent used when speaking the chosen language. Accents affect pronunciation, intonation, and speech patterns but not the language itself.
  Example - English (US) · English (UK) · English (Australia) · Spanish (Spain) · Spanish (Mexico)

​

​

Pronunciation Controls

This section defines how structured text is interpreted and spoken.

Word Replacements

Enable Word Replacements to substitute specific words or phrases before speech synthesis.

Use this to:

• Correct pronunciation of brand names
• Expand acronyms
• Standardize terminology

Example:

IVR → “interactive voice response”

Regex Replacement

Enable Regex Replacement to define pattern-based substitutions for structured text.

Use this for:

• Reference numbers
• Account IDs
• Alphanumeric codes

These rules are applied before synthesis to ensure clarity.

Preview Configuration

Use the Preview Configuration panel to test your settings before saving.

Text for Preview

• Enter custom text or select from predefined prompts such as:

  • Transaction summary
  • Upsell or offer
  • Greet the customer
  • Put the customer on hold

Generate Audio

1. Enter or select preview text.
2. Click Generate.
3. Listen to the output in the Audio Output section.

This helps validate pronunciation, tone, and overall voice experience.

Where This Configuration Is Used

Once created, the TTS configuration becomes available in Speech Profiles under:

Voice Care app → Voice Settings → Speech Profiles → TTS Source

• TTS configurations appear under User Configured

• Only configurations matching the selected language are shown

• These configurations are reused across voicebots and IVR flows

Important Notes

• TTS configurations are centrally managed and reused across deployments
• Changes to a configuration may impact all linked speech profiles
• Only configurations compatible with the selected language are displayed during usage

​

Availability Note
TTS Configuration is available only for new deployments created after 26.7 release. Existing deployments are not automatically migrated.

​

Related Articles

• Add Speech Profiles

• Introduction to Voice AI (STT and TTS)