Lex Mandarin Support

You are correct that Amazon Lex V2 supports Mandarin (PRC) and Cantonese (Hong Kong), but these languages are indeed not available in the Asia Pacific (Singapore) (ap-southeast-1) and Africa (Cape Town) (ap-south-1) Regions. This limitation applies to several other languages as well, as you've accurately listed.

Unfortunately, there is no publicly available information about a timeline for expanding language support in these specific regions. Amazon does not typically disclose future development plans or release dates for service expansions.

As for workarounds, there are a few options to consider, although they may not be ideal for all situations:

1. Use a different AWS region: If data residency requirements allow, you could potentially use Lex in a region where Mandarin is supported.

2. Implement a multi-region architecture: This could involve using Lex in a supported region for language processing, while keeping other parts of the application in the Singapore region. However, this approach would need careful consideration of data residency requirements and potential latency issues.

3. Use alternative services: Depending on your specific needs, you might be able to use other AWS services or third-party solutions that support Mandarin in the Singapore region.

4. Engage with AWS support: For critical business needs, you may want to engage directly with AWS support or your account manager to discuss your specific requirements and potential solutions.

It's important to note that AWS continuously works on expanding their services and capabilities, so it's worth keeping an eye on announcements for any updates regarding language support in these regions. However, as of now, there is no official timeline for when Mandarin support might become available in the Singapore region.
Sources
Languages and locales supported by Amazon Lex V2 - Amazon Lex
Latest features - Amazon Lex

We just launched an improved model for Mandarin (if you have an old bot, just rebuild it to use the updated improved speech-to-text model). I am looking into when we can have it in Sing.

I will push for this for the first months of next year.