Flow block in Amazon Connect: Set voice
This topic defines the flow block for setting the text-to-speech (TTS) language and voice to use for the contact flow.
Description
-
Sets the text-to-speech (TTS) language and voice to use for the contact flow.
-
The default voice is configured to Joanna (Conversational speaking style).
-
You can choose Override speaking style to make it and other voices Neural Voices. Neural voices make automated conversations sound more lifelike by improving the pitch, inflection, intonation, and tempo.
For a list of supported neural voices, see Neural Voices in the Amazon Polly Developer Guide.
-
After this block is run, any TTS invocation resolves to the neural or standard voice selected.
-
If this block is triggered during a chat conversation, the contact goes down the Success branch. It has no effect on the chat experience.
Supported channels
The following table lists how this block routes a contact who is using the specified channel.
| Channel | Supported? |
|---|---|
Voice |
Yes |
Chat |
No - Success branch |
Task |
No - Success branch |
Flow types
You can use this block in the following flow types:
-
All flows
Properties
The following image shows the Properties page of the Set voice block. It is configured for English, the voice is Joanna, and the speech style is Conversational.
Tip
For voices that support only neural speaking styles but not standard, the Override speaking style is automatically selected. You do not have the option to clear it.
You can also set language, voice, engine and style dynamically. There are a few configurations that must be followed when modifying the block:
-
If the language is selected dynamically, the voice must also be selected dynamically.
-
If the voice is selected dynamically and the speaking style is overridden, then the engine and style must be selected dynamically.
-
If the voice or engine are invalid, or the selected voice doesn’t support the selected engine, the error branch will be taken.
Note
-
Language code is only passed into a flow action if Set language attribute is selected. Therefore, invalid language codes will not take the error branch in this block but they may result in erroneous behavior when used with Lex V2 bots.
-
If a play prompt is added after the Error branch, the voice used for it will default to Joanna/standard.
-
If the defined speaking style is not supported by the defined voice, the None speaking style will be used.
-
Configuration
For a list of valid language codes, voices, and supported engines, see Available voices in the Amazon Polly Developer Guide .
Note
Amazon Connect supports standard and neural engines, so you can pass either standard or neural as values into the engine parameter.
To set the language attribute, pass in the specific language code into the parameter (for example, en-US or ar-AE). For the voice, simply pass the name of the voice (for example, Joanna or Hala).
Amazon Connect also supports speaking styles, which can be defined as None, Conversational, or Newscaster. The Newscaster and Conversational styles are both available for the following voices in the neural engine:
-
Matthew (en-US)
-
Joanna (en-US)
-
Lupe (es-US)
-
Amy (en-GB)
Note
If you don't specify an engine, the standard engine is used by default. However, some voices, such as Ruth (en-US), don't support the standard engine. For these voices, you must specify a supported engine. If you don't, the operation fails because Ruth doesn't support the standard engine.
The following table contains some examples on configurations and their results:
| Language Code | Voice | Engine | Speaking style | Result + Reasoning |
|---|---|---|---|---|
| en-US | Ruth | N/D | N/D | Error branch: engine is not specified, thus it defaults to standard. Ruth does not support standard engine, which results in error branch being taken. |
| en-US | Ruth | neural | none | Success branch: Ruth supports neural engine |
| en-US | Ruth | neural | conversational | Success branch: Even though Ruth does not support conversational speech style, the block does not take the error branch. Instead, when the voice is synthesized, it just uses no speaking style. |
| ar-AE | Ruth | neural | none | Success branch: This block does not do validation on language code. Only the voice is used to synthesize speech. However, language code being incorrect may result in erroneous behavior when used with Lex V2 bots. |
Use an Amazon Lex V2 bot with Amazon Connect
If you're using an Amazon Lex V2 bot, your language attribute in Amazon Connect must match the language model used to build your Lex bot. This is different than Amazon Lex (Classic).
-
If you build an Amazon Lex V2 bot with a different language model—for example, en_AU, fr_FR, es_ES, and more—under Voice, choose a voice that corresponds to that language, and then must choose Set language attribute, as shown in the following image.
-
If you're not using an en-US voice with an Amazon Lex V2 bot and don't choose Set language attribute, the Get customer input block results in an error.
-
For bots with multiple languages (for example, en_AU and en_GB) choose Set language attribute for one of the languages, as shown in the following image.
Configuration tips
-
For the Joanna and Matthew neural voices, in American English (en-US), you can also specify a Newscaster speaking style.
Configured block
The following image shows an example of what this block looks like when it is configured. It has the following branches: Success and Error.
Scenarios
See these topics for scenarios that use this block:
