Forum Discussion
AI Voice Generators
Got my AI Assistant today! Yay!
Hi! Our team is currently evaluating the AI Assistant features and so far, we’re pleased with them. The AI Assistant voices sound more natural; however, we have observed that the speaking rate can occasionally be somewhat rapid. I was informed that SSML tags can be used with the Standard out-of-the-box text-to-speech voices to adjust the speaking rate. And given that AI-generated voices are designed to naturally interpret word relationships, SSML tags cannot be applied to AI Assistant voices.
Are there any plans to enable the use of SSML tags with AI Assistant voices in the future?
We also found the need to (a) adjust the pace and (b) use SSML tags to adjust certain pronunciations when converting text to speech for our client eLearning courses, so we developed a tool using a Storyline front end that looks like this:
This allows us to cut and paste text (including SSML tags) into the text entry box, select a voice from the menu, adjust the pace and then, when the "Convert" button is clicked, we use a JavaScript trigger to pass the data over to a small node-js server-side app that calls Microsoft Azure AI Speech Services to do the text to speech conversion. The Storyline app reads out the text in the voice specified and offers the audio for download as an mp3 file.
Obviously you need an Azure account and a server to load the server-side app but it isn't that difficult to do. This produces a great result and saves us a heap of development time. We are pretty impressed with Microsoft's text-to-speech generative AI and, even though we use this all the time, we haven't yet exceeded our monthy free quota with Azure.
Thank you John!
