Behavior tags (old)
Behavior tags are supported for UneeQ Creator personas, and digital humans implemented after July 2020.
The behavior of your digital human can be controlled by sending specific instructions to the platform, known as behavior tags. You can add these tags to a dialogue response to determine the emotional state of your digital human when speaking the tagged dialogue.
Behavior tags are supported for both SSML (Speech Synthesis Markup Language) and plain-text (non-SSML) responses.
There are 6 behavior tags which are currently supported: <uneeq:happy>, <uneeq:empathetic>, <uneeq:excited>, <uneeq:joking>, <uneeq:agreeing> and <uneeq:enquiring>.
A positive emotional behavior which can be used for responses where warmth is required. This includes greetings.
<uneeq:happy>Good morning! What can I help you with today?</uneeq:happy>
Used for responses where a sense of caring and understanding should be displayed.
<uneeq:empathetic>I’m really sorry to hear that. I hope we can help resolve that issue for you.</uneeq:empathetic>
A more pronounced level of passion and energy will be expressed.
<uneeq:excited>Wow that is fascinating. I love to spend time outdoors as well!</uneeq:excited>
This is a cheeky expression which is often used when the digital human needs to tell a joke or when humour needs to be expressed.
<uneeq:joking>What do you call two monkeys that share an Amazon account? Prime mates.</uneeq:joking>
Used when the response indicates a sense of agreement with the user. This is useful for confirmations with the information the user has provided.
<uneeq:agreeing>Yes that’s a very good point. I see exactly what you mean.</uneeq:agreeing>
Requesting clarification, asking questions or where additional information is required.
<uneeq:enquiring>I couldn't understand what you said there, would you please be able to repeat that?</uneeq:enquiring>
Partially tagging a response
If you only want a behavior language tag to apply for part of your response, you must apply the <uneeq:neutral> tag to the rest of the response.
✔️ Correctly tagged partial response
❌ Incorrectly tagged partial response
Using multiple tags
You can also use multiple tags within a single response, so that your digital human can match appropriate behaviors to different words or sentences.
Best practices for using multiple behavior tags
- Avoid nesting behavior tags inside other behavior tags, as this will result in unexpected digital human behavior.
- We do not recommend using two different tags within a single sentence, as the transition between tags can introduce a slight delay to the Digital Human's speech, depending on your TTS provider.
Example of working with SSML
If you are using SSML, you'll need to ensure that behavior tags are included inside the <speak> tag.
⚠️ Note: Not all TTS providers support SSML, and most providers only support a subset of the tags defined in the SSML specification. We recommend following the SSML guidelines of your TTS provider.
You can combine <uneeq:behavior> tags with all SSML supported tags (refer to the list of supported tags by your TTS provider) to apply multiple effects on the digital human’s behavior and speech. Depending on your TTS provider, the order of the tags may be important.
When using Google TTS
If you are using Google TTS, ensure <uneeq:behavior> tags are not nested inside other SSML tags. Furthermore <uneeq:behavior> tags are required to directly follow speak tags.
✔️ Correct use of SSML with Google TTS
❌ Incorrect use of SSML with Google TTS
When using Microsoft Azure TTS
If you are using Azure TTS, ensure uneeq:behavior tags are not nested inside other SSML tags. Furthermore uneeq:behavior tags are required to directly follow speak tags.
✔️ Correct use of SSML with Azure TTS
❌ Incorrect use of SSML with Azure TTS
Other TTS providers
✔️ Correct use of SSML
✔️ Correct use of SSML