Before you record, I want to thank you deeply for assisting me with this project. Thank you! And I'll be happy to provide you with a score report from a native-speaker tutor if you include your email. Here is how I will process and use these recordings:

1. The recordings are used to train and evaluate the scoring model. They can also be heard by native-speaker tutors who assess them under confidentiality as well as those working on the project (currently only me) also under confidentiality. The recordings are never published, not ever shared beyond that, and not ever used to create a voice clone. Only aggregate results can ever be published.
2. They are stored under a codename, not your actual name. Your native language and JLPT level are stored with them. If you provide an email at the end, it is temporarily associated with this codename pending your results. Your codename will be included in the outgoing email that provides your scores. After 60 days or after an assessment has been provided for your recordings (whichever comes sooner), the association will be undone and your recordings will no longer have any identifier attached besides your codename, native language and JLPT level.
3. I do not expect that 60 days will pass without an assessment having been provided. If you gave an email and 60 days have passed without assessment, I will notify you of the circumstances and promptly destroy your recordings, which will never have been used for training (scoring from a tutor is required before training anyway).
4. Cloud storage and GPU compute in the EU (currently through Runpod) as well as software development tools may also process recordings. None of these providers may use the recordings for their own purposes.
5. No one working on this project or with access to these recordings will attempt to reidentify you. This, alongside the confidentiality clause, will be included in the agreement signed by the native-speaker tutors that will evaluate your pronunciation. You can have your data deleted from my set at any time, but you will need to email me at delete@relayed.me with your codename so I can delete the correct files within 30 days, including backups. Unfortunately, recordings already used to train a model cannot be removed from that model (in the form of statistical weights, not actual recordings), but after deletion they cannot be used in any later training.
6. If you stop partway, the words recorded so far are kept under these same terms. If the project changes hands, your recordings go with it under these terms or are deleted.
7. If you opt in to project updates, your email is kept on a separate list which is not linked to your codename or recordings, and you can unsubscribe at any time.

Enter your email to receive a native tutor pronunciation assessment on the words you recorded.

• This will temporarily be associated with your codename so I can send you the assessment, after which (or after 60 days, whichever comes sooner), the association will be undone.
• If you do not enter your email, I will not be able to send you your tutor-assessed scores, but I appreciate your participation nonetheless!