Privacy policy

We process account identifiers, profile and onboarding choices, practice activity and results, visual review cards, support messages, subscription status received from our merchant of record, security logs, and consent preferences.

When you request speech feedback, the submitted recording and derived pronunciation signals are processed for that request. When you ask to create practice from an image, short video, document, or note, that source is processed for the request; PDF, PPTX, DOCX, and EML files are converted to bounded extracted text before model processing. Mandarin Reflex stores a source hash and generated analysis, not the raw source. Professional scripts and evidence are stored in your account workspace when you use that feature.

We use this data to perform the service contract, authenticate accounts, deliver and improve requested practice, manage subscriptions and support, prevent abuse, and meet legal obligations. Optional performance measurement runs only with consent where consent is required.

Standard diagnostic audio and raw multimodal source files are processed for the requested result and are not retained by Mandarin Reflex after processing unless a feature explicitly offers retention. Generated material analyses expire after no more than 30 days and can be discarded before use. Review cards you choose to add remain in your account history until account deletion. Shadowing recordings are not uploaded or retained. Professional audio retention is off by default and, if enabled, follows the event-level deletion window you select.

Account and billing records are retained only as long as needed for the service, security, tax, dispute, and legal obligations. You can request an export, audio deletion, or account deletion from Settings; verified deletion jobs remove active account data, generated analyses, review cards, and private audio.

Mandarin Reflex uses Supabase for authentication, database, and private storage; Waffo as merchant of record; iFlytek for requested speech assessment; Alibaba Cloud Model Studio for requested Qwen language or multimodal generation; and MiniMax for reference speech. These providers process data under their service terms and applicable data-protection agreements.

Depending on your location, you may have rights to access, correct, export, delete, restrict, or object to processing and to withdraw consent. Use the in-product privacy controls or contact us. You may also complain to your local data-protection authority.

Controller: Wang Chao (王超), individual operator, Linfen, Shanxi, China. Privacy requests: support@mandarinreflex.com.