Privacy Policy — DreamVoice (Yume no Koe)
1. About This Policy
This Privacy Policy explains how the Operator handles user information in DreamVoice (“Service”). The Service is intended for users aged 13 or older. Users under 13 may not use the Service. If a higher age requirement applies under the applicable store age rating or local laws, that requirement applies. We do not knowingly collect information from users below the applicable age. If we learn that a user is under the applicable age, we will delete the information. Minors should use the Service with guardian consent.
New Today's Voice entries are no longer available in app version 3.14.0 and later. You can still read, delete, and use supported actions on saved records. References to Today's Voice in this Policy include existing records and use in older app versions.
2. Information Collected
The Service may collect the following:
- User input: dream text, Today's Voice text, optional titles, emotion tags, dream date / record date, notes or answers users write for daily Dream Voice or Today's Voice questions, and optional before-sleep notes and mood tags attached to a dream.
- AI outputs: daily Dream Voice analysis, supportive notes, symbols, summaries/keywords, Today's Voice feedback, Today's Voice hints/snapshots and safety support flags, and Weekly Dream Letter reflections, headlines, body text, highlights, prompts, symbols, and generated results that may reference or reflect source dream entries and notes for daily questions.
- Identifiers: anonymous auth ID (e.g., Firebase Auth UID), Support ID.
- Usage/technical data: screen/interaction events, compose mode changes, filter/segment taps, favorite/share/TTS taps, title edit sheet open/save/failure events (without title text), daily question note display/input/save/delete/sync-failure events (without note text or question text), Today's Voice generation/status events, Weekly Dream Letter display/create/ready/open/delete/push-open events, feedback recruitment prompt impression/open/dismissal events (without survey answers or contact details), crash logs, and config fetches.
- Dream Dictionary usage data: events for displaying the entry point, opening the Dictionary, searching, filtering, refreshing, navigating to a source dream, and correcting groupings (without individual symbol names, meanings returned for an individual dream, or dream text).
- Notification data: device tokens, notification permission status, and notification category settings needed to send notifications.
- Device and app information: OS version, app version, device model, language settings, Firebase Installation ID, IP address, and similar technical information.
- Apple Ads attribution information: whether an install was attributed to an ad, campaignId, adGroupId, keywordId, adId, country/region, supply placement, and similar fields. Fields such as keywordId are not collected when Apple does not return them.
- Server-stored data: user input (including dream text, Today's Voice text, and daily question notes), AI results, daily question note text, snapshots of the related question text, Today's Voice snapshots/hints/safety support metadata, character counts, timestamps, deletion/sync state, Weekly Dream Letter job state/results/source references, weekly aggregate information such as dream-entry counts, dates, emotion tags, keywords, and daily question note counts for the target week, processing state, usage counts.
- Dream Dictionary derived data: for each user, internal references to a completed Dream Voice and its source dream; dream date and language; symbols and meanings returned for that dream; links to a representative display symbol, variants, occurrence/display summaries, processing version/status, and user-requested grouping corrections (merge, split, or undo). This Policy refers to these items collectively as “Dream Dictionary derived data.”
- Contact and feedback survey information: for support inquiries, email address, inquiry content, and optional Support ID/device information/screenshots; for feedback surveys, selected answers, free-text responses, and optional contact details such as an email address.
The feedback survey does not ask users to provide their name, user ID, or dream content. The Service does not append a Support ID, anonymous authentication ID, dream identifier, dream content, or usage history to the survey URL, and does not link survey responses with in-app usage history.
Safety support flags, levels, and reason categories are used only to display supplemental in-app guidance when the input may include strong distress or safety concerns. They do not constitute medical, psychological, legal, or other professional judgment.
3. How It Is Collected
- Entered by users in the Service, support form, or feedback survey form.
- Automatically collected by the app (analytics, diagnostics).
- Issued via authentication (anonymous auth).
4. Purpose of Use
- Provide the Service (daily Dream Voice, Today's Voice, Weekly Dream Letter, display, storage, notifications).
- Store and display daily question notes and waking-life context, and use them for future Weekly Dream Letters.
- Generate, display, and update Your Dream Dictionary from completed Dream Voice outputs, and reflect user-requested grouping corrections.
- Maintain deletion integrity so a deleted dream is not shown in Your Dream Dictionary after a later sync.
- Manage usage limits, provide minimum safety support notices, and prevent abuse.
- Improve quality (troubleshooting, usage understanding).
- Measure Apple Ads performance and improve advertising operations.
- Understand Weekly Dream Letter generation status, reading behavior, and notification performance.
- Respond to inquiries and verify identity when needed (e.g., to prevent impersonation).
- Understand user feedback and requests and use them to improve the Service.
- Legal compliance and dispute response when required.
5. Third-Party Services
The Operator uses third-party services to provide and operate the Service. Their handling of information is governed by their own policies.
- Google Firebase (authentication, database, hosting, analytics, crash reporting, etc.)
- Apple AdServices (Apple Ads attribution)
- OpenAI (AI generation)
- Google Forms (support/inquiry and feedback survey forms)
Apple Ads Measurement
To measure Apple Ads performance, the attribution token obtained from Apple’s AdServices.framework is sent temporarily to Apple through the Service’s server. The token itself is not stored and is not written to logs or analytics. From Apple’s response, only attribution fields such as campaignId, adGroupId, and keywordId are stored on the device and recorded in Google Firebase Analytics. This process does not display the OS tracking-permission prompt.
Sending to OpenAI
The Service uses the OpenAI API for dream reflections, Weekly Dream Letters, and Dream Dictionary symbol organization. Content entered as text or transcribed from voice input is received by the Operator's server, and information needed for these features is sent to OpenAI over HTTPS. Recorded audio itself is not sent to OpenAI.
- Dream reflections: dream text, optional title, emotion tags, dream date, language, and optional intensity.
- Weekly Dream Letters: dream text, titles, dates, feelings, daily AI outputs, symbols, small prompts and your answers, and before-sleep events and mood. The input also includes the period, language, time zone, summaries of dream and Rest days, and references to source records. When Batch processing is used, the anonymous authentication ID and processing identifiers are also sent to OpenAI.
- Dream Dictionary symbol organization: symbol names and meanings returned for a newly completed Dream Voice, language, and candidate representative-symbol information. This classification request does not include dream text or a user identifier.
- Today's Voice in older app versions: entry text, optional title, emotion tags, record date, language, and the record identifier. New entries are no longer available in app version 3.14.0 and later.
Dream reflections and Weekly Dream Letters are processed on the server after you request generation. Related symbol organization and necessary retries or corrections may continue automatically. Generated text may also be sent for corrections. Saving an answer or before-sleep note alone does not start new AI generation. If your writing contains personal information such as names, that information is also sent. We do not guarantee that written content has been anonymized.
OpenAI Data Use, Protection, and Retention
The Operator requires AI service providers to provide the same or equal protection through purpose limitations, confidentiality, and security consistent with this Policy. We limit shared information to what is needed for these features and manage its handling in light of the provider's published conditions.
OpenAI states that API inputs and outputs are not used for model training or improvement by default. Abuse-monitoring logs are normally retained for up to 30 days, with longer retention possible for legal or safety reasons. Training use and retention for processing or monitoring are separate matters.
The Service disables storage of generated Responses API responses. This does not disable retention of monitoring logs, Batch files, or related data. OpenAI's published file specifications state that Batch input files expire after 30 days by default. Other related data may follow separate retention conditions.
For OpenAI's data-use and retention details, see the official “Data controls in the OpenAI platform” and “Upload file” documentation.
If You Do Not Want New Information Sent to AI
Do not start AI generation or retries; you can use draft saving and read saved records instead. This does not stop processing already started or retrieve information already sent. See “User Rights and Choices” for deletion within the Service. In-app deletion does not immediately delete OpenAI logs or files. Please contact the Operator with questions about data handling.
Your Dream Dictionary
Opening, searching, refreshing, or correcting the grouping in Your Dream Dictionary uses previously stored Dream Voice outputs and Dream Dictionary derived data on the Service's server. These actions alone do not send new dream text, symbols, or meanings returned for an individual dream to OpenAI for AI reanalysis.
Sending information to OpenAI when a new Dream Voice is generated, and when its symbols are classified after generation completes, is described above in “Sending to OpenAI.”
5-1. Official website analytics
On the official landing pages and dream-journal guides, we use Google Analytics to measure page visits and App Store link clicks for site improvements only after you allow it. The Google Analytics tag is not loaded before permission. You can use the site without allowing analytics.
We send an approved page URL without its query string or fragment, the page type, displayed language, a known traffic-source category, and the placement of an App Store link. We do not send arbitrary search terms, dream text, form input, or app account identifiers. Analytics is not started on pages with unknown query strings or URL fragments.
Google Analytics handles cookie-based browser identifiers and device and browser information. Google receives an IP address to handle the connection. Advertising cookies, advertising personalization, and Google Signals demographic and interest reporting are not enabled. Analytics cookies expire after 90 days.
Your choice and its timestamp are kept in this browser’s local storage for 180 days. You can change your choice through “Analytics settings” on each page. Withdrawing permission stops subsequent measurement, deletes this site’s analytics cookies, and reloads the page. It does not delete previously collected aggregates. Analytics is disabled when your browser signals Global Privacy Control or Do Not Track.
This choice applies to official website analytics. It is separate from the app’s analytics and crash reporting settings.
6. Disclosure to Third Parties
The Operator does not provide personal information to third parties without consent, unless required by law. However, data may be sent to the above service providers within the scope necessary to provide the Service.
7. Cross-Border Transfer
Because the Service uses cloud services provided by overseas operators, including Apple, Google Firebase, OpenAI, and Google Forms, collected data may be processed or stored on servers outside your country/region, including but not limited to the United States.
8. Retention
- On-device data: kept until the user deletes it.
- Server data: in principle kept until the user deletes it.
- Daily question notes: in principle kept until the user deletes them.
- Today's Voice results, snapshots, and job data: in principle kept until the user deletes them.
- Weekly Dream Letter results, fixed inputs, and job data in older storage (v1–v3): retained for restoration after a letter is deleted and hidden; subject to account deletion.
- Dream Dictionary derived data: kept for as long as needed to display and update the Dictionary and maintain deletion integrity. When an individual dream is deleted, content derived from it is excluded from the active Dictionary. To prevent it from reappearing after a later sync, source references, symbols, and meanings returned for that dream may be retained in a deleted state until the account is deleted.
- Contact and feedback survey information: kept as long as reasonably necessary to respond to inquiries or review and analyze feedback for Service improvements, and deleted when no longer needed.
- Logs, backups, audit records, abuse-prevention records, legal-compliance records, and dispute-response records held by third-party services: follow each provider’s retention rules or legally required periods and may persist for a certain period.
9. User Rights and Choices
- Delete individual dreams.
- Delete Today's Voice records.
- Delete daily question notes.
- Delete Weekly Dream Letters.
- Delete account (which removes related data on device and server).
- Push notification permissions: users can enable or disable push notifications through OS notification settings or in-app settings where available.
- Currently, there is no opt-out UI for the app’s analytics or crash reporting. For website choices, see “Official website analytics”.
Information submitted through the support form or feedback survey form is not automatically deleted when you use the in-app account deletion feature. If you want it deleted, please contact the Operator through the contact form and provide information that identifies the submission.
If you delete a Today's Voice record, the stored data, generated result, and related snapshot for that Today's Voice record are subject to deletion. Today's Voice is not used as a source for Weekly Dream Letter generation in v1.
Even if you delete an individual dream entry or daily question note, Weekly Dream Letters already generated before deletion may still contain summaries, keywords, symbols, reflections, or text based on that entry or note. Deleting a Weekly Dream Letter hides it from the list and retains its results for restoration; older input storage also retains fixed inputs. To remove related server data, use account deletion.
If you delete an individual dream entry, content in Your Dream Dictionary derived from that entry is excluded from the active Dictionary and is not shown again after later syncs. To prevent it from being shown again, source references, symbols, and meanings returned for that dream may be retained in a deleted state within your account. If you delete your account, Dream Dictionary derived data, including source-dream references, display summaries, and grouping corrections, is also subject to deletion.
Deletion mainly applies to data managed by the Service. Logs, backups, audit records, abuse-prevention records, legal-compliance records, and dispute-response records held by external services may remain for the period required by each service or applicable law.
Identifiers issued through anonymous authentication may remain on the device or be restored after reinstalling the app, depending on the device and OS behavior. To delete related server-side data, please use the account deletion feature.
10. Security
The Operator takes reasonable measures to prevent unauthorized access or leakage, including:
- Encrypted communication (HTTPS)
- Access control and authentication
- On-device data protected by standard OS mechanisms
11. Incident Response
If a breach or similar incident occurs, the Operator will investigate impact, prevent recurrence, and notify when legally required or deemed necessary.
12. Changes to This Policy
This Policy may be updated; notices will be given in-app or on the web.
13. Contact
For inquiries about this Policy: https://forms.gle/9eEJhVS64nKTxdgr8
Before-sleep notes attached to dreams
Additional information
We collect optional notes attached to a dream about the day’s events or thoughts before sleep (up to 200 characters), and before-sleep mood tags (up to three). Answers, notes, and mood tags can be saved independently. Missing entries do not mean unrelated or normal.
Storage and purpose
We store these personal records, their Dream references, timestamps, revision, and disclosure version on the server to save and display records and to generate future Weekly Dream Letters. When a letter is created, answers, waking-life events, and before-sleep states attached to its Dreams may be saved as fixed input and sent to OpenAI. Saving alone does not trigger AI generation or regenerate the original Dream Voice or an existing letter.
Analytics and history
Note text, answer text, and individual mood tags are not included in Analytics, Crashlytics, or ordinary logs. Original-record history, such as the calendar history, uses records currently available on this device. Weekly Dream Letter details check the referenced originals and adjust their display when edited, deleted, or unavailable for verification. Generated AI text stays as it was when created. Input counts come from the fixed generation input; they do not indicate how many records the AI prose used.
Deletion and retention
Answers, events, and before-sleep states can be deleted independently. Generated prose remains after the source record is edited or deleted. In older input storage (v1–v3), fixed inputs for a letter whose creation has begun may also remain. Deleted source fields are excluded from newly started generation inputs. Deleting a Weekly Dream Letter hides it from the list and retains generated prose for restoration; older input storage also retains fixed inputs. To remove related server data, use account deletion. Account deletion includes these records and their weekly fixed inputs.
Replies after a gentle question
Additional replies and your follow-up answers
Saving an answer to a gentle question stores its text, Dream and question references, revision and timestamps on your device and our server. Saving alone does not request new OpenAI generation. Choosing “Receive a reply” sends that dream’s text, feelings, saved first answer, original question, daily interpretation and display language to OpenAI to generate a reply and an optional follow-up question. It also sends the dream date, the date and time the Dream Voice was requested, and the date and time the answer was last saved when available. Your answer to that question is saved independently. Saving it does not generate an AI closing message.
Use in Weekly Dream Letters
Starting a letter for the Dream’s original week captures its first answer, all follow-up answers and before-sleep records saved on the server at capture time. Edit history for the same question is not included more than once. AI questions remain separate from your answers. Later saves do not automatically update an existing letter or carry into another week.
Retention and deletion in the new input storage
The new input storage (v4) keeps frozen dream text, answers and before-sleep records in private temporary data while processing, and deletes that data on completion, failure or replacement. Deleting a related personal record also removes active fixed input; checks stop further sends and publication from that input. The fallback expiry is seven days for weekly input and 24 hours for additional-reply input. Physical deletion exactly at expiry is not guaranteed. Identifiers, revisions, digests, counts and deletion state may remain for acceptance, limits, duplicate prevention and deletion consistency. Account deletion removes related data.
Saved prose and external services
Deleting the first answer hides its additional reply and AI question, which may quote it, and starts cleanup; your independently saved follow-up answers remain. Generated Weekly Dream Letter prose is not rewritten automatically. Weekly fixed inputs from older storage (v1–v3) follow the existing retention terms. Calendar history shows personal records currently available on this device, once per Dream. New-feature Analytics events exclude text, questions, Dream IDs and actual dates. Deletion cannot atomically recall an external send already confirmed before it. OpenAI monitoring logs, Batch files and related data have separate retention terms; immediate retrieval through in-app deletion is not guaranteed.