Issues with Mixed Chinese-English or Chinese-Numeric Scenarios

#49
by jiuface - opened

Description

In scenarios where Chinese-English or Chinese-numeric text is mixed, certain anomalies occur. For example:

  1. Numbers are not pronounced correctly.
  2. Some words or phrases are skipped or mispronounced.

Steps to Reproduce:

1.Input a sentence containing both Chinese and English, such as "今天是Monday".
2. Input a sentence with Chinese and numbers, such as "今天是2025年".

Expected Behavior:

  1. Numbers and mixed languages should be pronounced accurately.

Observed Behavior:

  1. Numbers are mispronounced or skipped.
  2. Mixed language phrases result in incorrect or incomplete pronunciation.

Question:

Are there any existing solutions or recommendations to address these issues?

Hi @jiuface ,
Thanks for reporting this issue. This looks like it's likely an issue with the model itself, so please consider opening an issue on the GitHub repo.
Thanks!

mrfakename changed discussion status to closed

Sign up or log in to comment