International Phonetic Alphabet

The International Phonetic Alphabet (IPA) is a pronunciation notation that writes down every sound humans can make using the same symbols, regardless of language. If you write a foreign word using another language's script, different people will read it slightly differently — but if you write it in IPA, speakers of any language can produce exactly the same sound. Just as musical notation fixes pitch with precise symbols, IPA fixes pronunciation with symbols.

In speech synthesis AI, this notation serves as a practical tool. For example, when an AI mispronounces a word like "subpoena," whose spelling gives no clue how it should be read, developers can specify the exact pronunciation directly using IPA symbols. This works like building a pronunciation dictionary for words the AI is prone to get wrong — company names, acronyms, unfamiliar personal names, and so on.

Cartesia's speech synthesis model Sonic-3.6 supports a pronunciation dictionary feature that lets developers specify a word's pronunciation directly with IPA symbols, such as writing subpoena as "<<s|ə|ˈ|p|i|n|ə>>". This is a practical feature meant to prevent the AI from mispronouncing specific words, used in settings like call centers where accurate pronunciation matters.

1. Enter an English word into an online IPA converter and check its phonetic transcription.
2. Look up a foreign word you've often found confusing (e.g., a brand name) in IPA and compare it to how a native speaker actually pronounces it.
3. If you use a speech synthesis service, check whether it offers a pronunciation dictionary feature, and try specifying a word's pronunciation using IPA notation.