Supported Languages
The languages supported by AmiVoice API are as follows. The table also summarizes the availability of features for each language.
| Language | Word Registration | Pronunciation Registration Method | Automatic Punctuation Insertion |
|---|---|---|---|
| Japanese | Available | Hiragana/Katakana | Supported |
| English | Not Available | - | Supported |
| Chinese | Available | Pinyin (Note) | Supported |
| Korean | Available | Hangul | Supported |
Word Registration/Pronunciation Registration Method
The ability for users to register unrecognized vocabulary and the method for setting pronunciations differ for each language. The English engine does not support word registration. For details, please see Word Registration.
Automatic Punctuation Insertion
Punctuation marks are automatically inserted into the speech recognition result text. This feature uses a neural network trained on a large amount of data to insert punctuation marks where deemed appropriate based on the context. The inserted punctuation marks differ for each language.
| Language | Automatically Inserted Characters |
|---|---|
| Japanese | 、 。 ？ |
| English | . , ? |
| Chinese | 。 ？ 、 ， |
| Korean | . , ? |
For Korean, spacing (띄어쓰기) is applied according to the context.
English
Number Expressions
For English numbers, single-digit numbers are output as English words, while numbers with two or more digits are converted to Arabic numerals.
Chinese
About Pronunciation Registration
Chinese pronunciations should be described in Pinyin with tones represented by numbers. For example, "我们" should be written as "wo3men5".