Privacy Policy

Yimberzol Foundation

Last updated: July 2026 Applies to: yimberzol.org and record.yimberzol.org Operator: Yimberzol Foundation, Anantnag, Jammu & Kashmir, India

This Privacy Policy ("Policy") is issued by Yimberzol Foundation ("the Foundation," "we," "us," or "our"), a not-for-profit company incorporated under Section 8 of the Companies Act, 2013. The Foundation operates yimberzol.org (the "Website") and record.yimberzol.org (the "Platform"), the voice data collection application for Project Zoon, an open Kashmiri speech dataset initiative (the "Project").

This Policy describes the categories of personal data we collect, the purposes for which it is processed, the circumstances under which it is published, and the rights available to you. By accessing the Platform and submitting recordings, you consent to the collection, processing, and publication of your data as described in this Policy and in the Contributor Terms of Service.

1. Scope

This Policy applies to:

• Visitors to yimberzol.org, to whom Section 2 applies; and
• Contributors who submit voice recordings through record.yimberzol.org, to whom this Policy applies in its entirety, in particular Sections 3 through 9.

2. The Website (yimberzol.org)

The Website is informational. No account creation or submission of personal information is required to browse it.

Our hosting and infrastructure providers may automatically log standard technical information — including IP address, browser type, pages accessed, and timestamps — for security, diagnostic, and operational purposes. This information is not used to identify individual visitors and is not sold or otherwise monetized.

Correspondence sent to the Foundation by email or WhatsApp is retained for the purpose of responding to and maintaining a record of that correspondence.

3. The Platform (record.yimberzol.org)

3.1 Categories of data collected

When you contribute to the Project, the Foundation collects the following categories of data:

(a) Identity and account information — retained internally, never published:

• Full name;
• Contact information (email address and/or telephone number);
• Authentication details provided by your sign-in provider (for example, the email address associated with your Google account, where Google sign-in is used).

(b) Contribution data — published as part of the dataset:

• Voice recordings and the corresponding text of the sentences read;
• A pseudonymous speaker identifier ("Speaker ID") assigned to you by the Foundation.

(c) Demographic and technical metadata — published as part of the dataset:

• Age or age range;
• Gender;
• Accent and/or dialect;
• Region;
• Recording device information (such as device category or microphone type);
• Recording environment conditions (such as indoor/outdoor setting and ambient noise characteristics);
• Any other non-identifying attributes requested during onboarding.

3.2 Data that will never be published

Your name and contact information will not, under any circumstances, be included in the published dataset. This information is retained internally in access-restricted systems, and is used exclusively to:

• associate your contributions with you for the purpose of honoring withdrawal requests;
• contact you in connection with the Project where necessary; and
• maintain auditable records of consent.

4. Nature of Voice Data

Voice recordings are inherently capable of identifying a person. Even in the absence of a name, a voice constitutes a biometric characteristic, and published demographic metadata (age, gender, accent, region) may, in combination, narrow the range of possible speakers — particularly within small communities. The Foundation pseudonymizes all published data through Speaker IDs; however, the Foundation cannot warrant that re-identification of a speaker from voice data alone is impossible.

You should contribute only if you are comfortable with your voice, in your accent and dialect, being publicly and permanently available under an open license.

5. Use and Publication of Data

5.1 Purpose of processing

Data is collected and processed for the purpose of building, curating, and publishing an open Kashmiri speech dataset for the training and evaluation of speech and language technologies, including automatic speech recognition (ASR), text-to-speech (TTS), and related systems.

5.2 Review and curation

Following each recording session, submitted recordings undergo review, quality assessment, annotation, and curation by the Project's reviewers. This review process represents a material investment of reviewer and community effort in each accepted recording, and is the operational basis for the withdrawal terms set out in Section 6.

5.3 Publication

Recordings, transcripts, Speaker IDs, and the metadata described in Section 3.1(b) and (c) will be published as a public dataset — including on dataset hosting platforms such as Hugging Face — under a dataset license to be determined by the Foundation.

The terms of the dataset license, including any conditions or restrictions on commercial use, rest at the sole discretion of the Foundation and have not yet been finalized. The Foundation intends the dataset to be openly and freely available for research, community, and non-exploitative use, and reserves the right to impose conditions on certain categories of use — including large-scale commercial use — to protect the interests of the Project and its volunteer contributors. The applicable license will be published alongside the dataset and referenced on the Platform prior to publication.

Depending on the final license terms, third parties worldwide — potentially including commercial entities — may download, use, modify, and redistribute the dataset in accordance with those terms. Upon publication, the dataset ceases to be under the Foundation's exclusive control.

5.4 Pre-publication holding period

The Foundation will not publish any contributor's recordings earlier than fifteen (15) days after the recording session in which they were made. This period constitutes the contributor's guaranteed window of withdrawal, as set out in Section 6.1.

6. Withdrawal, Deletion, and Contributor Rights

6.1 Guaranteed withdrawal — within 15 days of recording

For a period of fifteen (15) days following each recording session, you may request the complete deletion of the recordings made in that session, together with their associated data, by contacting the Foundation through the channels in Section 12. The Foundation will honor such requests unconditionally.

6.2 Requests after the 15-day period

Upon expiry of the 15-day period, accepted recordings will typically have undergone review, quality assessment, and annotation by Project reviewers, and may have been integrated into curated dataset releases. Deletion at this stage would forfeit the reviewer and community effort already invested in the material.

Accordingly, once fifteen (15) days have elapsed from the date of a recording, the Foundation does not guarantee deletion of that recording, whether or not it has yet been published. Removal requests received after this period will be considered in good faith, but the decision to delete rests at the sole discretion of the Foundation.

Where the Foundation does elect to remove published data, removal is limited to what lies within its direct control: the Foundation can remove recordings and metadata from its own dataset versions and distribution channels and exclude them from future releases. Because the dataset is openly licensed and publicly downloadable, the Foundation cannot retract copies, mirrors, forks, or derivative works — including trained models — already obtained or created by third parties. Post-publication removal is therefore necessarily incomplete.

Contributors should treat the 15-day period described in Sections 5.4 and 6.1 as their effective opportunity to withdraw.

6.3 Internal-use-only arrangements

Where mutually agreed in writing through the Foundation's official email (info@yimberzol.org), the Foundation will retain a contributor's recordings and associated data for internal use only and exclude them from the published dataset. Any such arrangement must be documented prior to publication of the data concerned. Where such an arrangement exists, it supersedes the publication provisions of this Policy in respect of that contributor's data (see Section 11).

6.4 Access and correction

You may request information regarding the data the Foundation holds about you, and request correction of inaccurate account information or metadata, through the channels in Section 12.

7. Eligibility and Minors

Contribution through the Platform is restricted to persons eighteen (18) years of age or older. The Foundation does not knowingly collect recordings through the Platform from any person under 18. If the Foundation becomes aware that a contributor is under 18, the contributor's recordings and account information will be deleted.

Participation by minors, where undertaken, is conducted exclusively through separate, offline sessions with documented and verifiable consent from a parent or legal guardian, in accordance with applicable law, including the Digital Personal Data Protection Act, 2023.

8. Disclosure to Third Parties

The Foundation engages third-party infrastructure providers for the operation of the Website and Platform, including hosting, deployment, database storage, and authentication services (for example, web hosting providers, cloud database providers, and sign-in providers such as Google). These providers process data solely to the extent necessary to deliver their services and are not authorized to use it for their own purposes.

Publication of the dataset itself, as described in Section 5, is distinct from this operational disclosure.

The Foundation does not sell personal information.

9. Data Retention

• Name and contact information: retained internally for the duration of the Project to honor withdrawal requests, maintain consent records, and enable necessary contact.
• Published recordings and metadata: persist as part of the public dataset indefinitely, subject to Section 6.
• Recordings subject to a deletion request made within 15 days of recording: deleted (Section 6.1).
• Recordings subject to a removal request made after 15 days: retained or deleted at the Foundation's discretion (Section 6.2).
• Recordings governed by an internal-use-only arrangement: retained internally and excluded from publication (Section 6.3).

10. Security

Internal contributor information — names, contact details, and consent records — is stored in access-restricted systems, with access limited to personnel who require it for the Project. The Foundation applies reasonable technical and organizational measures to protect the data under its control. No system is entirely secure, and the Foundation cannot guarantee absolute security, but it undertakes to minimize risk to the information it holds.

11. Superseding Agreements and Amendments

Superseding agreements. Any agreement concluded with a contributor through official channels — email, a signed or written consent form, or other documented communication — supersedes this Policy to the extent of any conflict.

Amendments. The Foundation may amend this Policy from time to time. The "Last updated" date at the head of this Policy reflects the current version. Material amendments affecting data already collected will, where reasonably practicable, be communicated through the contact channel you have provided. Continued use of the Platform following the effective date of an amendment constitutes acceptance of the amended Policy.

12. Contact and Grievance Redressal

For withdrawal requests, access or correction requests, grievances, or any other matter concerning this Policy or your data:

• Email: info@yimberzol.org
• WhatsApp: +91 96225 42946

The Foundation endeavors to respond to all data-related requests within a reasonable time.

This Policy applies to all visitors to yimberzol.org and all contributors onboarded through record.yimberzol.org, operated by Yimberzol Foundation. It operates alongside the Contributor Terms of Service and the Project Zoon Dataset License.