TTS Generator

Convert text to natural-sounding speech using AI-powered text-to-speech synthesis. Generated audio is automatically saved to your Audio Library for immediate use in campaigns and IVR flows.

Overview

The TTS (Text-to-Speech) Generator transforms written text into spoken audio files. It supports multiple Indian languages and speaker voices, making it ideal for creating localized voice content without recording studio time, voice actors, or audio editing software.

Supported Languages and Speakers

Sarvam AI (Indian Languages)

Deepgram Aura (English)

Use Cases

Welcome Messages and Greetings

Generate personalized welcome messages for new users. Use TTS to dynamically insert names, company names, or other variables into voice prompts without recording individual files.

Example: “Hello [Customer Name], welcome to Acme Services. Your account has been activated successfully.”

Notifications and Alerts

Create timely audio notifications for transactional events — payment confirmations, order status updates, appointment reminders, or service alerts.

Example: “Your electricity bill payment of rupees two thousand five hundred was received on July 15th. Thank you.”

Multi-Language Campaigns

Build campaigns targeting audiences across different regions by generating the same message in multiple languages. Upload all variants to the Audio Library and segment your campaign contact lists by language preference.

Example workflow:

1. Write your message in English.
2. Translate and generate TTS variants in Hindi, Tamil, Telugu, and Bengali.
3. Segment contacts by preferred language.
4. Launch four parallel campaigns, each using the appropriate language audio file.

Rapid Prototyping for IVR Flows

During IVR flow development, use TTS to quickly generate placeholder prompts for testing. Once the flow logic is finalized, replace TTS-generated prompts with professionally recorded audio if desired.

Dynamic Content

For campaigns where message content changes frequently (e.g., daily deals, stock prices, weather alerts), TTS eliminates the need to re-record audio for every update.

Generating TTS

Step-by-Step

1. Navigate to Audio Library from the sidebar.
2. Click Generate Voice.
3. Select the Target Language from the dropdown.
4. Choose a Speaker Voice — preview a sample by clicking the play icon next to each speaker name.
5. Enter your Text in the text area. Use the formatting tips below for best results.
6. Click Generate Voice. Processing typically takes 3–10 seconds depending on text length.
7. The generated audio file appears in your Audio Library with a name matching the first few words of your text. You can rename it for clarity.

Text Formatting Tips for Better TTS Output

The quality of TTS output depends significantly on how you structure your input text. Follow these guidelines for the best results:

• Use proper punctuation. Periods, commas, and question marks create natural pauses and intonation. A well-punctuated sentence sounds far more natural than a run-on block of text.
• Spell out acronyms and abbreviations. Write “United States” instead of “U.S.”, “Doctor” instead of “Dr.”, “January 15th” instead of “15/01”. The TTS engine interprets these phonetically and may mispronounce abbreviations.
• Write numbers as words for critical values. “Your balance is rupees two thousand five hundred” sounds clearer than “Your balance is rupees 2500.00” for financial amounts. For phone numbers, write digit by digit: “nine eight seven six five four three two one zero”.
• Use line breaks to create pauses. Each new line or paragraph adds a brief pause in the generated speech, helping to separate logical sections of your message.
• Keep sentences under 20 words. Longer sentences strain the TTS engine’s prosody model and can lead to unnatural rhythm.
• Avoid special characters. Symbols like @, #, $, %, &, *, and emojis are either mispronounced or produce garbled output. Write out the words they represent.
• Specify pronunciation for ambiguous words. For words with multiple pronunciations (e.g., “lead” as in metal vs. guidance, “read” as present vs. past tense), the engine picks the most common. Test and adjust phrasing if the output is incorrect.

Combining TTS with Audio Files in Campaigns

TTS-generated audio and pre-recorded audio files can be used together in a single campaign for a polished, professional result:

Troubleshooting

Issue: TTS generation fails with an error.

Cause: API configuration issue — the TTS engine requires a valid AI provider API key.

Solution: Verify that your OpenAI or Gemini API key is configured and active in Settings . Test the key by generating a short TTS sample.

Issue: Generated audio sounds robotic or unnatural.

Cause: Poorly formatted input text — run-on sentences, missing punctuation, or excessive special characters.

Solution: Review the Text Formatting Tips section above. Break long sentences into shorter ones, add punctuation, and remove special characters. Regenerate with corrected text.

Issue: TTS mispronounces names or technical terms.

Cause: The TTS engine uses phonetic rules that may not match your intended pronunciation.

Solution: Try phonetic spelling (write the word as it sounds). For example, if the name “Siobhan” is mispronounced, try writing “Shivon” instead. Test multiple spelling variations until the output is correct.

Issue: Text is truncated in the generated audio.

Cause: The input text exceeds the 1,000-character limit.

Solution: Split the text into chunks of 1,000 characters or fewer. Generate each chunk separately, then combine the audio files. Ensure each chunk ends at a natural sentence boundary.

Next Steps: If the issue persists after these checks, visit the Support page with the error message and the text you attempted to generate.

Related Articles

• Audio Library
• Voice Campaigns
• IVR Flow Builder
• Voice Features Overview