Skip to main content

​
Overview

Pronunciation dictionaries allow you to fine-tune how our text-to-speech (TTS) engine pronounces specific words. If you find that a word, such as a company name, a technical term, or a proper noun, is consistently mispronounced, you can add it to a pronunciation dictionary to provide a custom, text-based phonetic spelling.

​
How It Works

The feature works by mapping a word to its desired pronunciation using simple text. When the TTS engine encounters a word in your dictionary, it will use your specified pronunciation instead of its default. For example, if the word “Atoms” is being pronounced incorrectly, you could provide the following mapping:
  • Word: Atoms
  • Pronunciation: A-toms
This tells the TTS engine to pronounce “Atoms” as “A-toms”.

​
Current Limitations

  • Text-to-Text Only: The pronunciation must be provided as plain text.
  • No IPA Support: Currently, we do not support the International Phonetic Alphabet (IPA) for specifying pronunciations. You’ll need to use a textual representation that guides the TTS engine to the correct sound.
By using pronunciation dictionaries, you can improve the clarity and accuracy of your voice agent, ensuring a better experience for your users.