Forum Discussion
Generate Phonemes using AI and a Dictionary
Pronunciation is a challenge at times, especially when the same word is pronounced differently depending on its "use."
We use a combination of IPA phonemes and basic phonetics (phonics). I use an AI chat to generate the SSML code, then place it in a SharePoint library for when I need it again. The same voice may get it right one time, but not the next, so I never know when I'll need the phoneme. Given that the Articulate voices don't allow emphasis on syllables, this is the best method I've found so far. If anyone has a better way, it would be most appreciated if you would share, thank you!
This is very basic, but it does work when needed. Here is an example of the AI prompt: Please generate the phoneme SSML code for the word "<word>" and use https://www.merriam-webster.com/dictionary to obtain the phonemes.
Example for "use" as a noun:
<speak>
<phoneme alphabet="ipa" ph="ˈjuːs">use</phoneme>
</speak>
Example for "use" as a verb:
<speak>
<phoneme alphabet="ipa" ph="ˈjuːz">use</phoneme>
</speak>
2 Replies
Thank you for this best practice tip! But where exactly should this piece of code be inserted? I tried using it with an example sentence during speech creation (AI-assistant):
How to
<speak>
<phoneme alphabet="ipa" ph="ˈjuːz">use</phoneme>
</speak>
the text-to-speech functionality. Click on...
But the word "use" was not spoken at all. Some words were created, but it was neither "use" nor the SSML code.However, the closed captions were exactly right.
Please allow me to provide additional information. Some companies did not opt to purchase the AI add-on, so we must work a bit differently. We use Copilot to produce SSML code needed using the IPA phoneme standard. The built-in AI may already serve that purpose, so you may not need the code.
Either way, when Insert Audio is selected, the code is placed directly into the Text-to-Speech block. The phoneme is placed wherever the word appears in the text.
