Online API for Hanzi-to-Pinyin conversion

[westmeadboy]

I've seen a few threads about Hanzi-to-Pinyin conversion but they talk about various websites rather than APIs that can be accessed using something like JSON.

I know you can access Google Translate using JSON and that Google Translate provides pinyin romanization of Hanzi, but is it possible to access that through JSON also?

Alternatively I could just build my own offline converter based on CC-CEDICT data. It wouldn't be perfect but probably good enough for my needs. In cases where a character maps to multiple pinyins I would just use the most common one.

How does that sound?

[lingjoin]

In cases where a character maps to multiple pinyins I would just use the most common one. --------> The problem lies in such cases are very common. The precision would be depressive.

A better process recommendated is:

1. Making word segmentation on input sentences. (Unlike English, a Chinese sentence should be splitted into a token sequence.)

2. Map each word to pinyin. The map is often unique. The technique is state-of-the-art. And we have collect such mapping list before.

[roddy]

Have a look at Adsotrans - if they're not doing it, I doubt anyone else is.

[renzhe]

Yeah, I think that adsotrans does exactly this, and should be considerably more accurate than a CEDICT-based character-for-character conversion.