AWS Polly Pinyin doesn't say certain things. Is there a document with valid Pinyin?

Question

Take for example `貪小便宜`. If I put this text into the console, it correctly speaks it. `tan xiao bian yi`. However if I input SSML as follows: `貪小便宜`, the voice only says `tan pian yi`, it skips the `xiao`. I know certain Pinyin can be romanized in different ways, but `xiao` seems to be pretty standard from what i've seen. I've also tried `xiao1-4` none of which worked.

Is there a way to get a list of the valid pinyin for Polly so I can normalize my pinyin to use what is accepted by Polly?
I know the voice can say what I'm looking for, as it works when I put the chinese characters in, however for my usecase they won't always be as readily available as the Pinyin, but the Pinyin seems to be less consistent.

Accepted Answer

Nevermind. As is always the case, I spent a long time looking for answers before posting, and minutes after posting I find a solution.

The `phoneme` tag needs to be split up. So it needs to be sent to Polly as:

```

This behavior is different from the standard IPA which does seem to accept arbitrary lengths of IPA under a single `phoneme` tag.

AWS Polly Pinyin doesn't say certain things. Is there a document with valid Pinyin?

Relevant content