Choose a voice for Twilio ConversationRelay

To review the list of voices from Google, see the Twilio TTS Voices documentation.

To sample an Google Text-to-Speech voice and set your default voice:

1. Go to the Text-to-Speech page in the Twilio Console.
2. Choose Google Text-to-Speech from the Your Default Provider dropdown menu.
3. Click the Test & Configure link. The Test & Configure Your Default Voice modal appears.
4. Choose a language and locale from the Language dropdown menu.
5. Choose a voice from the Voice dropdown menu.
6. If desired, change the text in the Test this voice prompt. Any voice tests read the text in this box.
7. Click the play button (▶) next to the Test this voice prompt.
8. When you fing your desired voice, click Save. This closes the Test & Configure Your Default Voice modal.

The default voice for your ConversationRelay changed to your chosen voice and the Default Voice option displays your chosen voice.

1. Browse the available voices in the Available voices and languages table on the Text-to-Speech page.
2. To filter the available choices to Google only, set Provider to Google.
3. You can filter the available voices using these parameters:
   • Language and locale
   • Type
   • Gender
4. Test them using the Twilio Console to find the one that best fits your application's requirements.
5. Select the Voice ID in the Voice column in the table. An example would be en-US-Wavenet-D.
6. Copy this value.

1. Create or open an existing TwiML file.

2. Create a new <Connect> tag or go to an existing one.

3. In the <Connect> tag, add an unpaired <ConversationRelay> noun.

4. In the <ConversationRelay> noun, set the following attributes:

   • Add a url attribute set to the value of your Websocket server.
   • Add a ttsProvider attribute set to the value of Google.
   • Add a voice attribute set to the value of the voice ID copied in the previous section.

   Example TwiML app that uses Google for ConversationRelay

   Copy code block

   1
   <Connect>
   2
     <ConversationRelay 
   3
       url="wss://example.com/websocket"
   4
       ttsProvider="Google"
   5
       voice="en-US-Chirp3-HD-Charon"
   6
       ... />
   7
   </Connect>

5. Save your TwiML file.

To review the list of voices from Amazon Polly, see the Twilio TTS Voices documentation.

To sample an Amazon Polly voice and set your default voice:

1. Go to the Text-to-Speech page in the Twilio Console.
2. Choose Amazon Polly from the Your Default Provider dropdown menu.
3. Click the Test & Configure link. The Test & Configure Your Default Voice modal appears.
4. Choose a language and locale from the Language dropdown menu.
5. Choose a voice from the Voice dropdown menu.
6. If desired, change the text in the Test this voice prompt. Any voice tests read the text in this box.
7. Click the play button (▶) next to the Test this voice prompt.
8. When you fing your desired voice, click Save. This closes the Test & Configure Your Default Voice modal.

The default voice for your ConversationRelay changed to your chosen voice and the Default Voice option displays your chosen voice.

1. Browse the available voices in the Available voices and languages table.
2. To filter the available choices to Amazon Polly only, set Provider to Polly.
3. You can filter the available choices using these parameters:
   • Language and locale
   • Type
   • Gender
4. Highlight the Voice ID in the Voice column in the table. An example would be Joey-Neural. Don't copy the * in that column. That symbol identifies a language as bilingual.
5. Copy this value.

1. Create or open an existing TwiML file.

2. Create a new <Connect> tag or go to an existing one.

3. In the <Connect> tag, add an unpaired <ConversationRelay> noun.

4. In the <ConversationRelay> noun, set the following attributes:

   • Add a url attribute set to the value of your Websocket server.

   • Add a ttsProvider attribute set to the value of Polly.

   • Add a voice attribute set to the value of the voice ID copied in the previous section.

     Example TwiML app that uses Amazon Polly for ConversationRelay

     Copy code block

     1
     <Connect>
     2
       <ConversationRelay 
     3
         url="wss://example.com/websocket"
     4
         ttsProvider="Polly"
     5
         voice="Joey-Neural"
     6
         ... />
     7
     </Connect>

     To learn more about the <ConversationRelay> noun, see the <ConversationRelay> documentation.

5. Save your TwiML file.

ElevenLabs defaults to their Flash 2.5 model.

1. Browse the list of ElevenLabs voices.

   View the list of ElevenLabs voices

2. Find a voice that matches your requirements. You can filter the list by language, accent, voice quality (category), age, gender, and any tags applied to the language.

3. Click the play button (▶) next to the voice row.

1. From the table, click Copy VoiceID in the row of your chosen voice.

2. Create or open an existing TwiML file.

3. Create a new <Connect> tag or go to an existing one.

4. In the <Connect> tag, add an unpaired <ConversationRelay> noun.

5. In the <ConversationRelay> noun, set the following attributes:

   • Add a url attribute set to the value of your Websocket server.
   • Add a ttsProvider set to the value of "ElevenLabs".
   • Add a voice attribute set to the value of the copied VoiceID.
     • Add one or both optional settings for the voice attribute.
       • Add an audio model. The voices from ElevenLabs default to the Flash 2.5 model. You can choose other models and could improve the quality or performance of your application depending on your use case. You can use a different model by appending a hyphen to the VoiceID followed by the model ID. The supported model IDs include flash_v2, turbo_v2_5, turbo_v2, and the default, flash_v2_5. Some models only work with a specific set of languages. To learn about the strengths and the supported languages of each model, see the ElevenLabs website.
       • Add the speed, stability, and similarity of your chosen ElevenLabs voice. At the end of your selected voice or audio model value, add a hyphen, then follow it with an underscore-separated string with values for speed, stability, and similarity in that order. You can't set only one of these values. If you set one, you must set them all.
         • The speed value should be a value between 0.7 and 1.2.
         • The stability and similarity values can range from 0.0 to 1.0.

   For example: A voice value of XrExE9yKIg1WjnnlVkGX-1.2_0.6_0.8 sets its speed to 1.2, its stability to 0.6, and its similarity to 0.8.

   Setting	Type	Necessity	Accepted values	Separator
   Voice ID	String	Required	20 alphanumeric character identifier from ElevenLabs	-
   Audio Model	String	Optional	flash_v2, flash_v2.5, turbo_v2, turbo_v2.5	_
   Speed	Number	Optional	0.7 to 1.2 inclusive	_
   Stability	Number	Optional	0.0 to 1.0 inclusive	_
   Similarity	Number	Optional	0.0 to 1.0 inclusive	_

   To learn more about how these settings affect your application's voice, see [the ElevenLabs documentation][].

6. Save your TwiML file.

Example TwiML app that uses ElevenLabs for ConversationRelay

1
<Connect>
2
  <ConversationRelay 
3
    url="wss://example.com/websocket"
4
    ttsProvider="ElevenLabs"
5
    voice="NYC9WEgkq1u4jiqBseQ9-turbo_v2_5-0.8_0.8_0.6"
6
    ... />
7
</Connect>