Privacy Policy · Murmur 喃喃
Last updated: November 18, 2026
TL;DR
- Murmur runs on your device. Your voice, paintings and transcripts are created, stored and replayed on the device you use.
- To provide the Service and keep it running well, your session data — including voice recordings, transcripts and inferred emotion categories — may be transmitted to our secure infrastructure and processed there under appropriate technical and organisational safeguards. Wherever practicable, analytical processing is carried out in de-identified, pseudonymised or aggregated form. Your voice is used to produce your painting and to improve the Service, not to identify you.
- We do not sell your data, share it with advertisers, or use it to train publicly available AI models.
- In the iOS app, your recording and its transcript are sent to our backend (hosted on Google Firebase) and to OpenAI's API, which converts your speech to text and writes your reflection prompt — but only after you tap Agree on the consent screen shown during onboarding. If you do not agree, nothing is recorded or sent. In the web version, two optional paths may exist instead — your browser's built-in speech recognition and, if you supply your own key, AI-assisted emotion analysis. Each is described below.
- You are always free to clear your data, stop using an optional feature, or delete Murmur.
1. Who we are
Murmur ("we", "us", "our") is a personal voice diary application that transforms one minute of your voice into a painting. This Privacy Policy explains what personal data may be processed when you use Murmur, on what legal basis, and what rights you have. If you have questions, contact us at [email protected].
2. Scope
This policy applies to your use of the Murmur application, whether accessed as a mobile app, a progressive web app, or a hosted webpage. It describes personal data associated with your use of the Service; it does not cover any third-party website, service, or platform to which you may be redirected.
3. Categories of personal data
The following categories of personal data may be processed in connection with the Service. Whether a particular category is processed depends on how you use the Service and which optional features you enable.
| Category | Description | | :--- | :--- | | Voice input | The audio captured by your device's microphone during a Murmur session. May be transmitted to our infrastructure for the purposes described in section 4, and retained for no longer than necessary for those purposes. | | Transcripts | The textual transcription of your session, generated by your device, the browser you use, or — in the iOS app — OpenAI's speech-to-text service. | | Emotion inferences | Categorised signals (e.g., "joy", "calm", "gratitude") derived from your voice and transcript in order to render your painting. | | Paintings and metadata | The image generated for each session, together with related metadata such as its palette, timeline, and name. | | Account data | In the iOS app, when you use Sign in with Apple, we receive a user identifier and — if you choose to share it — your name and email address. Authentication is handled by Firebase Authentication (Google LLC). | | Preference data | Language preference, feature toggles, and any API credentials you choose to store. | | Diagnostic data | Anonymous or aggregated information about how the Service performs (e.g., crash counts, session length), where such information is collected at all. |
We do not knowingly process contacts, location, advertising identifiers, browsing history, or data belonging to other applications on your device.
4. How your data is processed
Murmur is designed to keep personal data on your device by default. Personal data may be processed in the following ways:
- On-device processing — Voice capture, prosodic analysis, keyword-based emotion scoring, painting generation, and the storage of paintings and preferences all take place on the device on which you use the Service. Data processed this way does not reach us, and we do not have access to it.
- Your device's or browser's speech recognition service — Where the Service uses your operating system's or web browser's built-in speech recognition to produce a transcript, that service may transmit audio to the vendor of that operating system or browser in order to return text. The behaviour, retention, and further uses of such data are governed by the vendor's own privacy policy (for example, Google's or Apple's), which you can consult through the settings of your device or browser.
- AI processing in the iOS app (OpenAI) — In the iOS app, after you tap Agree on the consent screen shown during onboarding, your voice recording and transcript are transmitted to OpenAI, L.L.C. (United States) through its API to (a) convert your speech to text and (b) generate a short reflection prompt for your entry. OpenAI processes API requests under its own terms and privacy policy and, under its API data-usage policy at the time of writing, does not use API content to train its models. If you do not agree, the app does not record and does not transmit any audio or text. You can withdraw your consent at any time in the app's settings; the app will then stop sending data for new sessions.
- Optional AI-assisted analysis (web version only) — If, and only if, you supply your own AI service credentials (for example, an Anthropic API key) in the web version's settings, short transcript excerpts may be transmitted to the corresponding AI provider (currently Anthropic, PBC, United States) to obtain additional emotion signals. The provider processes the request under its own terms and privacy policy, and returns results to your device. You may remove your credentials at any time to disable this feature.
- Server-side processing to provide, maintain, and improve the Service — To provide, maintain, secure, and improve the Service, your session data — including voice recordings, transcripts, prosodic features, and inferred emotion signals — may be transmitted to infrastructure operated by us or by service providers acting on our behalf (for the iOS app: Google Firebase — Firebase Authentication, Cloud Firestore and Cloud Storage, operated by Google LLC), and processed there under appropriate technical and organisational safeguards. Data is encrypted in transit and is stored under reasonable access controls. Wherever practicable, analytical processing that is not required to render the Service back to you personally is carried out in de-identified, pseudonymised, or aggregated form, with direct identifiers separated from analytical records where feasible. Your voice is used to produce transcripts, prosodic features and the resulting painting, and to help us maintain and improve the Service; it is not used to uniquely identify you (see section 12).
- Cloud storage, where applicable — Some deployments of the Service may offer optional cloud synchronisation of your library. Where such a feature is offered and enabled by you, encrypted copies of the relevant data may be stored on cloud infrastructure operated by us or by a service provider on our behalf.
5. Legal basis (EEA / UK / other regions with GDPR-equivalent laws)
We rely on the following legal bases under Article 6 and, where applicable, Article 9 of the General Data Protection Regulation:
| Processing | Legal basis | | :--- | :--- | | Providing the core Service (recording, transcription, on-device emotion analysis, painting generation, local storage, playback) | Performance of a contract with you — Article 6(1)(b) | | Optional AI-assisted analysis that you enable by supplying your own credentials | Your explicit consent — Article 6(1)(a); and, to the extent emotion signals may qualify as data revealing information about your health or psychological state, Article 9(2)(a) | | De-identified, pseudonymised, or aggregated server-side analysis used to maintain, secure and improve the Service (see section 4) | Our legitimate interest in operating and improving a reliable Service, balanced against your rights and freedoms — Article 6(1)(f) | | Anonymous or aggregated diagnostic information used to maintain and improve the Service | Our legitimate interest in operating a reliable Service — Article 6(1)(f) | | Communications required to respond to your enquiries or to comply with legal obligations | Legitimate interest — Article 6(1)(f); or legal obligation — Article 6(1)(c) |
You may withdraw any consent you have given at any time; withdrawal does not affect the lawfulness of processing carried out before withdrawal.
6. Purposes
Personal data described in section 3 may be processed only for the following purposes:
- To operate the Service you have chosen to use.
- To render, store and replay the outputs of a session (paintings, transcripts, gratitude entries) on your device.
- To generate optional personalised features (such as historical views, weekly summaries or signature palettes) from data already present on your device.
- To debug, maintain, secure and improve the Service, where possible using anonymous or aggregated information.
- To respond to enquiries about the Service, and to satisfy legal, regulatory or law-enforcement obligations to which we are subject.
We do not use personal data for automated advertising, for the training of publicly available AI models, or for any purpose that would produce legal effects concerning you or similarly significantly affect you (see section 12).
7. Third parties and international transfers
We do not sell or rent your personal data. Where personal data is disclosed to a third party in the course of providing the Service, such disclosure is limited to what is necessary and, where relevant, is governed by data-processing terms with that third party. Third parties that may process personal data in connection with the Service include:
- The vendor of the speech recognition service used by your device or browser (for example, Google when the Service is used through the Chrome browser). Any transmission of audio to such a vendor is initiated by that vendor's own component of your operating system or browser, is subject to its privacy policy, and is not routed through us.
- OpenAI, L.L.C. (United States) — in the iOS app, performs speech-to-text and generates reflection prompts from your recording and transcript, after your consent (see section 4). OpenAI processes API requests under its own terms and privacy policy and, at the time of writing, does not use API content to train its models.
- Google LLC (Firebase) (United States) — provides authentication (Firebase Authentication, including Sign in with Apple), database (Cloud Firestore) and file storage (Cloud Storage) for the iOS app's backend.
- AI service providers you have chosen to use in the web version by supplying your own credentials, currently Anthropic, PBC (United States) for its Claude family of models. Anthropic processes API requests in accordance with its published terms and privacy policy and, at the time of writing, does not use API-submitted content to train its publicly available models.
- Apple, Inc. — for app distribution, Sign in with Apple, and, if applicable, in-app purchases.
- Cloud-infrastructure and analytics service providers engaged by us to host the infrastructure on which the server-side analysis described in section 4 is performed, and to receive de-identified or aggregated diagnostic data. Such providers are bound by contract to process personal data only on our documented instructions and to apply appropriate security measures.
Some of these third parties are located outside the European Economic Area, the United Kingdom, or your country of residence, including in the United States. Where personal data is transferred outside your jurisdiction, we rely on appropriate safeguards recognised by applicable law — such as the European Commission's Standard Contractual Clauses, your explicit consent, or other legally recognised transfer mechanisms.
8. Storage location and security
Personal data processed on your device is stored on that device, under the security controls of the operating system and browser you use. Where any personal data is transmitted to a third party (see section 7), such transmission is protected in transit by industry-standard encryption (typically TLS 1.2 or above), consistent with what the third party makes available. Where we (or a service provider acting on our behalf) hold personal data on servers, we apply reasonable and appropriate organisational and technical measures, having regard to the nature of the data and the risks concerned.
No security measure is perfect. If we become aware of a personal data breach that is likely to result in a risk to your rights and freedoms, we will notify you and, where required, the competent supervisory authority in accordance with applicable law.
9. Retention
We hold personal data for no longer than is necessary for the purposes described in this policy or as required by applicable law. In practice, this means:
- Voice input, transcripts, prosodic features, and inferred emotion signals transmitted to our infrastructure are retained for no longer than necessary for the purposes described in section 4, and may be earlier deleted, aggregated, or de-identified in line with our retention practices.
- Transcripts, emotion inferences, paintings and preferences remain on your device for as long as you keep them, and are removed when you clear the Service's data from within your device settings, uninstall the application, or otherwise instruct removal.
- De-identified, pseudonymised, or aggregated records used for the server-side analysis described in section 4 may be retained for the periods necessary to maintain and improve the Service. Where such records can no longer reasonably be linked back to you, they may be retained indefinitely for statistical purposes.
- Anonymous or aggregated diagnostic data may be retained for longer periods for statistical purposes, provided it cannot reasonably be linked back to you.
You may request further information about our retention practices, or the deletion of any personal data we hold about you, by contacting us at the address in section 15.
10. Your rights
Subject to the conditions and exceptions set out in applicable law, you have the following rights in relation to personal data concerning you:
- Access — to obtain confirmation of whether personal data concerning you is processed and, if so, a copy of that data.
- Rectification — to have inaccurate personal data corrected.
- Erasure — to have personal data erased where the conditions of Article 17 GDPR (or an equivalent local right) apply.
- Restriction — to obtain restriction of processing in the cases set out in Article 18 GDPR (or its equivalent).
- Portability — to receive personal data you have provided in a structured, commonly used, machine-readable format, where processing is based on consent or on a contract with you and is carried out by automated means.
- Objection — to object, on grounds relating to your particular situation, to processing based on our legitimate interests.
- Withdrawal of consent — to withdraw at any time consent you have previously given, without affecting the lawfulness of processing based on consent before its withdrawal.
- Complaint — to lodge a complaint with a competent supervisory authority in your country of residence, place of work, or place of an alleged infringement.
Where processing takes place on your device, many of these rights can be exercised directly through the Service's own settings or through your device's application-management controls. Otherwise, please contact us at the address in section 15.
If you are in Taiwan, the equivalent rights under Article 3 of the Personal Data Protection Act apply, including the right to enquire, request review, request a copy, request supplementation or correction, request cessation of collection, processing or use, and request deletion.
11. Third-party content in your sessions
You may choose to speak about other individuals during a session. Where a transcript references identifiable third parties, that content is treated in the same way as your other transcripts (see sections 4 and 9). You are responsible for the lawfulness of any personal data of third parties that you introduce into the Service, particularly if you choose to share the resulting painting, transcript or other output with others.
12. Automated decision-making
Emotion inferences produced by the Service are used to generate a painting and related personalised features. They are not used to make decisions that produce legal effects concerning you, or that similarly significantly affect you, within the meaning of Article 22 of the GDPR. Any interpretation of emotion signals is offered as a creative aid, not as a diagnostic or clinical assessment.
The Service does not perform biometric identification. Voice input is processed for the purpose of transcription and expressive analysis only; it is not used to uniquely identify any individual.
13. Children
Murmur is not intended for children under the age of 13, and, in the European Union, not intended for individuals below the age of digital consent applicable in their country of residence (typically 16, but as low as 13 in some Member States). We do not knowingly process personal data of children below the applicable age. If you believe a child has provided us with personal data, please contact us and we will take appropriate steps to remove such data.
14. Changes
We may update this Privacy Policy from time to time. Where a change materially affects how personal data is processed, we will provide reasonable notice through the Service and, where practicable, other appropriate channels before the change takes effect. Where an update requires your consent under applicable law, we will obtain that consent before applying the change to you.
15. Contact
For questions about this Privacy Policy, to exercise any of the rights described in section 10, or to raise any concern about the processing of your personal data, please contact us at:
- Email: [email protected]