Terms of Service

Dataset.ET is an open speech and language data initiative operated by Snapwre Technologies PLC, a company registered in Addis Ababa, Ethiopia (“we”, “us”, the “Project”).

These Terms cover the website at dataset.et, the Dataset.ET Telegram bot, the datasets we publish, and any email we send you. Our Privacy Policy explains what personal data we handle and forms part of these Terms.

By using the site or the bot you accept these Terms. If you do not accept them, please do not use them.

Contributors read prompts aloud through the Telegram bot and may validate other people's recordings. Participation is voluntary and you may stop at any time.

You must be 18 or older to contribute, and you must only submit recordings of your own voice. Do not submit audio containing another person's voice, audio you did not record yourself, or audio containing personal information about anyone.

By submitting a recording you grant us a perpetual, worldwide, royalty-free licence to include it in openly published datasets, including under Creative Commons Attribution 4.0 (CC BY 4.0). Published recordings are pseudonymised: your Telegram identity and phone number are never included in a release.

Because open datasets are copied and mirrored by others, a recording that has been published cannot be recalled from everyone who has downloaded it. We will remove your data from our systems and from future releases on request; we cannot undo distribution that has already happened.

Contributors may earn credits for accepted recordings and for validation work, redeemable for mobile airtime or other payouts we offer from time to time. Rates, caps and eligibility are set in our platform settings and may change.

Credits are not currency, carry no cash value until redeemed through the mechanisms we provide, and are not transferable. We may withhold or reverse credits obtained through duplicate submissions, automated activity, recordings that are not your own voice, or validation that does not reflect a genuine assessment.

Datasets we publish are released under the licence stated on the release itself, normally CC BY 4.0. You may use them commercially and for model training, provided you give attribution as that licence requires.

Datasets are provided as is. We document what we measured and what we did not, and we make no warranty that a dataset is fit for any particular purpose.

You must not attempt to re-identify contributors, to link pseudonymous identifiers back to individuals, or to combine our releases with other data for that purpose.

Our directory lists datasets published by other people. We do not host, control, endorse or validate them, and each is governed by its own licence and its publisher's terms. Verify both before using anything you find there.

If you submit a dataset for listing, you confirm that you may lawfully share the link and the information you provide. We review every submission by hand and may decline or remove any listing at our discretion.

We send only transactional email: replies to a message you sent us, notifications about a submission or account action you took, and operational notices about work you have done on the platform. We do not send marketing email, and we do not sell, rent or share email addresses with anyone for their own marketing.

We only email an address that a person gave us for that purpose, or that we are replying to. We do not send to purchased, harvested or scraped lists.

Every message identifies Dataset.ET as sender and gives a working reply address. To stop receiving email from us, reply to any message or write to [email protected], and we will remove you.

We monitor delivery. Addresses that hard-bounce are suppressed immediately and not retried, and any address that reports a complaint is suppressed permanently and never contacted again.

You must not:

• submit recordings that are not your own voice, or that you obtained from elsewhere
• submit the same audio against multiple prompts, or otherwise attempt to inflate credits
• use automated tools to record, validate or redeem on the platform
• operate more than one contributor account, or attempt to circumvent payout caps
• submit unlawful, hateful, or knowingly false content, or content infringing anyone's rights
• attempt to gain unauthorised access to our systems, or disrupt their operation

We may suspend or terminate access, withhold unredeemed credits, and exclude affected recordings from releases where these Terms are breached.

The site, the bot and the datasets are provided without any guarantee of availability. We may change, suspend or discontinue any part of the Project, and we may update these Terms. Material changes will be reflected in the effective date above and, where the change affects contributors, announced through the bot.

To the fullest extent permitted by Ethiopian law, we are not liable for indirect or consequential loss arising from use of the site, the bot or the datasets, including loss of data, profit, or the results of any model trained on our data.

Nothing in these Terms limits liability that cannot lawfully be limited.

These Terms are governed by the laws of the Federal Democratic Republic of Ethiopia, and disputes fall to the courts of Addis Ababa.