Mandarin Pronunciation Study

You will be asked for your language background information and your voice will be recorded and used to build speech recognition models that can help Mandarin language learners. You will be able to choose how your speech is used. One option is to restrict the recordings to researchers at the University of St. Thomas and immediate collaborators. Another option is to share your recordings more broadly through a gated (non-public) repository (HuggingFace) with researchers for educational purposes only. Hugging Face is a model and data repository that promotes the use of shared data and models. By sharing your data here, we hope to encourage research in the area of pronunciation assessment by making our project transparent and repeatable.

No identifying information besides voice will be shared and only self-hosted speech recognition tools will be used (i.e., your speech will not be processed by third party services, only locally running software and models). That being said, your speech is considered personally identifying information, so you can choose whether your speech will be shared as a gated, nonpublic Hugging Face dataset, or restricted just to our research group.