search

ibm-cloud-docs / text-to-speech

text-to-speech
6 stars 26 forks source link

Clarify that phonemes must be supported by the custom model language #129

Closed Trinobanto closed 4 years ago

Trinobanto commented 4 years ago

When adding entries to a custom voice model using the phonetic method, it should be more clear than the phonemes provided should be supported by the custom voice model language.

For instance, when synthesizing a text in Spanish that contains an English word, one can think that a good approach to pronounce that word correctly is to use the pronunciation method with that word in English; then, take the phonetic response and add the word/translation pair to the Spanish custom model. However, you get error saying that ipa characters are not supported, etc.

jeffpk62 commented 4 years ago

Hi, Victor. I saw this only recently. I have added a new section to address this shortcoming. Thanks for taking the time to post the recommendation!