Who Speak Exactly is for
Speak Exactly is designed for college-level learning and personal speaking practice. It is not offered to children under 13. Do not create an account or use Speak Exactly if you are under 13. A professor or institution may set additional rules for a course.
What you provide and what Speak Exactly creates
Speak Exactly stores account details, your legal and preferred names, course enrollments, uploaded documents and recordings, written and spoken responses, transcripts, generated questions and feedback, practice history, and progress information when those features are used. Some independent practice saves your work and progress automatically; saving a rehearsal recording is a separate choice. Your professor may also provide course materials, assignments, grades or checkpoint observations.
How AI is used
Speak Exactly sends the material needed for an AI feature to OpenAI through Speak Exactly’s API connection. This may include your topic, document text, speech script, audio or transcript, responses, and relevant course source excerpts. OpenAI returns questions, transcription, guidance, drafts, or feedback. Speak Exactly may store those results in your account. OpenAI states that API data is not used to train its models by default unless the API customer opts in. OpenAI may retain limited data for abuse monitoring or legal reasons under its current policies. Speak Exactly cannot promise that any online AI service has zero retention or risk.
Who can see coursework and practice
Your professor and authorized course staff can see work you submit for their course, including the responses, feedback, and recordings made available by that activity. Your independent practice and private Study Studio are not automatically shared with your professors. Authorized Speak Exactly operators and service providers may access information needed to operate, secure, and support the service. Other students do not gain access to your private work merely because they share a class.
Voice setup, recordings, and speaker matching
Before course work opens, Speak Exactly asks for three voice reference samples: reading aloud, familiar conversation, and thinking aloud. You can listen and retake them before final submission. Submitted reference samples remain saved to your account and cannot be replaced or deleted through the student interface. Contact Speak Exactly about a data-removal request. They are for comparison and learning context; they do not by themselves prove identity or misconduct. Automatic speaker matching with a separate provider is not currently active. If Speak Exactly later enables that optional provider, it will explain the provider and request a separate choice before sending a sample there. For an assigned spoken activity, the recording and its transcript may be saved and available for professor review. Independent rehearsal recordings are saved when you choose to save them.
Feedback, progress, and academic decisions
Speak Exactly can show recall, response timing, speaking habits, course completion, and evidence of demonstrated understanding over time. AI feedback can be mistaken, and speech patterns, accents, pauses, fillers, or a changed speaking style are not proof of cheating. Speak Exactly does not claim to detect AI-written work or make academic-integrity decisions. Professors are responsible for their own review and course decisions. Course completion points track completed tasks; they are separate from demonstrated understanding.
Files and sensitive work
Speak Exactly may preserve an uploaded original file for viewing or download while extracting text or audio for learning features. Only upload material you have a right to use. A syllabus, student notes, research draft, or unpublished thesis may contain confidential information or third-party rights. Check your institution’s, advisor’s, or publisher’s rules before submitting such material. Remove personal details that a learning task does not need.
Service providers, security, and retention
Speak Exactly uses Supabase for accounts, application records and live files; Cloudflare for site delivery and private backup copies of uploaded files and recordings; OpenAI for AI features; and an email provider for account messages. Data is sent as needed for the feature used. Private file snapshots follow a 30-day backup cycle. Database backups and other providers have separate retention arrangements; the file-snapshot period is not a promise about every provider’s copies. Access controls and security measures reduce risk, but no online service can guarantee absolute security.
Private data, exports, and removal requests
Private practice, uploaded private files, saved recordings and voice setup samples remain while your account is active. Speak Exactly does not automatically delete them at semester end or because of inactivity. After a verified removal request is approved, eligible live private data is scheduled for removal within 30 days. A removal case blocks further account activity while it is fulfilled. An export can include your profile, responses, conversations, practice histories, released grades and feedback, your group contributions, associated live files and your own teaching sources. Other people’s work, staff-only unpublished evaluations, provider-held copies and some shared content require a separate review before release. An export is not itself a deletion. Contact [email protected] for a verified request.
Coursework, shared work, and departing professors
Submitted coursework, grades and shared/group records are reviewed separately under the responsible professor’s or institution’s documented retention direction. Speak Exactly records which categories remain, why they remain and when that decision must be reviewed. Private data may be removed while reviewed academic records remain. Other students’ work is not erased because one person leaves. Academic identity is considered during that review rather than blindly removed from every record. Before removing a professor’s account, the owner must transfer its courses and linked teaching recordings to an authorized professor, or explicitly approve archiving and retaining the courses. Removing sign-in does not automatically erase retained course records.
Backups, recovery, and privacy receipts
Removing a live file does not immediately erase existing private backup snapshots; ordinary file copies expire under the 30-day cycle. Exceptional purges require a scoped owner review so other accounts’ files are preserved. Speak Exactly keeps a separate removal/suppression record and checks it during recovery to avoid reviving removed data. Privacy fulfillment receipts contain request identifiers, verification/action dates, removed or retained categories, retention reasons and an operator reference; they do not retain exported responses or recordings. Completed receipts and their associated suppression records are kept for 12 months, unless an explicit documented hold applies. Historical backups require a separate review before restoration. Submitted voice setup removal also removes that setup; continuing required spoken work later requires a new setup or an approved accommodation.
Choices, changes, and contact
You can turn off optional notifications and sound effects in Settings. You can choose whether to save some independent practice recordings. You may stop using an AI activity instead of submitting new material. This agreement is required to use Speak Exactly’s AI-powered learning and course features; declining it means those features will not open. If the agreement changes in a meaningful way, Speak Exactly will show a new version and ask you to read and accept it before continuing. For privacy, account, or data requests, contact [email protected].
Provider information: OpenAI API data controls ↗.