Emotional TTS with Amazon Polly's SSML tags

Question

Polly allows the use of SSML tags which is great, but I cannot seem to find a tag that would add some kinds emotion. Let's refer to this as Emotional TTS. For instance, the following snippet:

<speak> Hi! My name is Joshua. </speak>

<speak> Hoi! Mijn naam is Lotte. <break time="3s" /> Ik lees elke tekst voor die je hier invoert. </speak>

Is it possible to do Emotional TTS with Polly; e.g. "aaaws" and "aaahs" with Polly? I guess if it were, it would have to be via SSML. Right? How would I modify the above to facilitate such a thing?

Answer 1

If you are using Alexa, depending on language, you might be able to achieve some of this with speechcons:

<speak>
    Not quite "aah", but almost:
    <say-as interpret-as="interjection">argh</say-as>.
</speak>

See the supported list of speechcons and more details on: https://developer.amazon.com/en-GB/docs/alexa/custom-skills/speechcon-reference-interjections-english-us.html

Unfortunately, Amazon Polly does not support interjections as of August 2021. You can try it interactively in the browser with the code above, which gives me:

The input you provided isn't valid. Check the input, and then try again.