Text-to-Speech Configuration

Text-to-Speech (TTS) converts written text into spoken audio during call flows. Instead of pre-recording messages, you can type the text and have the system generate speech automatically. This is especially useful for dynamic messages that change frequently or include personalized data.

How TTS Works in Exotel

TTS is used within the Greeting applet and IVR applet:

1. You enter text in the applet's TTS configuration
2. When a call reaches the applet, Exotel's TTS engine converts the text to audio in real time
3. The generated audio is played to the caller
4. The call flow continues to the next applet

Text input ("Welcome to Acme Corp") ──► TTS Engine ──► Audio played to caller

Available Languages and Voices

Indian Languages

Language	Language Code	Female Voice	Male Voice
English (India)	en-IN	Yes	Yes
Hindi	hi-IN	Yes	Yes
Tamil	ta-IN	Yes	--
Telugu	te-IN	Yes	--
Kannada	kn-IN	Yes	--
Malayalam	ml-IN	Yes	--
Bengali	bn-IN	Yes	--
Marathi	mr-IN	Yes	--
Gujarati	gu-IN	Yes	--
Punjabi	pa-IN	Yes	--

International Languages

Language	Language Code	Female Voice	Male Voice
English (US)	en-US	Yes	Yes
English (UK)	en-GB	Yes	Yes
Arabic	ar-XA	Yes	Yes
Malay	ms-MY	Yes	--

info

Language and voice availability may vary. Check the Exotel dashboard for the most current list of supported languages and voices for your account.

Configuring TTS in Applets

Basic TTS Setup

1. In the Flow Builder, click on a Greeting or IVR applet
2. Select Text-to-Speech as the audio source
3. Enter the text you want spoken
4. Select the Language from the dropdown
5. Select the Voice (female or male)
6. Click Preview to hear the generated speech
7. Click Apply to save

Configuration Options

Setting	Description	Default
Language	The language for speech generation	English (India)
Voice	The voice persona (female/male)	Female
Speed	Speech rate (slow, normal, fast)	Normal
Pitch	Voice pitch (low, normal, high)	Normal

SSML Support (Enterprise)

Speech Synthesis Markup Language (SSML) provides fine-grained control over how text is spoken. Enterprise accounts can use SSML tags to customize pronunciation, pauses, emphasis, and more.

Supported SSML Tags

Tag	Purpose	Example
<speak>	Root element (required)	<speak>Hello</speak>
<break>	Insert a pause	<break time="500ms"/>
<emphasis>	Emphasize a word	<emphasis level="strong">important</emphasis>
<prosody>	Control rate, pitch, volume	<prosody rate="slow">Please listen</prosody>
<say-as>	Control how text is interpreted	<say-as interpret-as="telephone">9876543210</say-as>
<sub>	Substitute pronunciation	<sub alias="Exotel">EXOTL</sub>
<p>	Paragraph (adds natural pause)	<p>First paragraph.</p><p>Second.</p>
<s>	Sentence (adds brief pause)	<s>First sentence.</s><s>Second.</s>

SSML Examples

Reading a Phone Number

<speak>
  Your reference number is
  <say-as interpret-as="telephone">4412345678</say-as>.
</speak>

The TTS engine reads each digit individually instead of treating it as a large number.

Adding Pauses

<speak>
  Welcome to Acme Corporation.
  <break time="1s"/>
  Please select from the following options.
  <break time="500ms"/>
  Press 1 for Sales.
  <break time="300ms"/>
  Press 2 for Support.
</speak>

Controlling Speed and Pitch

<speak>
  <prosody rate="slow" pitch="+10%">
    Your order has been confirmed.
  </prosody>
  <break time="500ms"/>
  <prosody rate="medium">
    The delivery will arrive by tomorrow.
  </prosody>
</speak>

Spelling Out Text

<speak>
  Your booking code is
  <say-as interpret-as="characters">ABCD1234</say-as>.
</speak>

Reading Currency

<speak>
  Your account balance is
  <say-as interpret-as="currency">INR 5000.50</say-as>.
</speak>

Prosody Attributes

Attribute	Values	Effect
rate	x-slow, slow, medium, fast, x-fast, or percentage (e.g., +20%)	Speech speed
pitch	x-low, low, medium, high, x-high, or semitones (e.g., +2st)	Voice pitch
volume	silent, x-soft, soft, medium, loud, x-loud, or dB (e.g., +6dB)	Speech volume

Say-As Interpret Types

Type	Description	Input Example	Spoken As
cardinal	Number as quantity	123	"one hundred twenty three"
ordinal	Number as ordinal	3	"third"
characters	Spell out each character	ABC	"A B C"
telephone	Phone number format	9876543210	"nine eight seven six five four three two one zero"
date	Date reading	2024-01-15	"January fifteenth, twenty twenty-four"
currency	Currency amount	INR 500	"five hundred Indian rupees"

Dynamic TTS with Passthru

Combine TTS with the Passthru applet to generate personalized messages:

Passthru (fetch customer data) ──► Greeting (TTS: "Hello {CustomerName}, your order {OrderID} is on the way")

Your server returns the customer name and order ID, which are inserted into the TTS text before speech generation.

TTS Best Practices

1. Keep messages concise -- Long TTS messages sound monotonous; break them into shorter segments
2. Use natural phrasing -- Write text as you would speak it, not as you would write it
3. Add pauses -- Use <break> tags between sections for a more natural feel
4. Spell out abbreviations -- Write "rupees" instead of "INR" unless using SSML say-as
5. Test with real callers -- Listen to TTS output over a phone line, not just in preview
6. Consider pre-recorded audio -- For critical customer-facing greetings, professional voice recordings sound better than TTS
7. Use appropriate language -- Match the TTS language to your caller demographics

Troubleshooting

Issue	Possible Cause	Resolution
TTS not playing	Applet not configured correctly	Verify text and language settings
Wrong pronunciation	Language mismatch	Select the correct language for the text
Numbers read incorrectly	TTS interprets as quantity	Use SSML say-as to control interpretation
Robotic sound quality	Default TTS engine	Consider upgrading to premium voices (Enterprise)
SSML not working	SSML not enabled on account	Contact account manager to enable SSML support

Related Topics

• Greeting Applet Guide -- Using TTS in greetings
• IVR Applet Guide -- TTS for IVR prompts
• Passthru Applet Guide -- Dynamic data for TTS
• Flow Builder Guide -- Building flows with TTS applets