Data License

Version 2026-11-11.

This license sets out what a buyer of an otoSpeech data pack may do with the data.

1. Parties

This Data License is between otoearth, Inc., a Delaware corporation ("oto"), and the organization or individual named as licensee at checkout ("Licensee").

The person who accepts this license at checkout confirms that they may bind the Licensee.

2. Definitions

• "Dataset" means the audio files and metadata of the otoSpeech pack the Licensee bought, and any update oto delivers for that pack.
• "Derived Data" means transcripts, labels, features, embeddings, and other data made from the Dataset.
• "Model" means a machine learning model trained, fine-tuned, or evaluated with the Dataset or Derived Data.
• "Removal Notice" means an email from oto that lists sessions to be removed from the Dataset.
• "Public Sessions" means sessions that oto has also published under the Creative Commons Attribution 4.0 license in its public otoSpeech releases.

3. License

When payment is complete, oto grants the Licensee a non-exclusive, non-transferable, worldwide license, with no right to sublicense, to:

• store and copy the Dataset inside the Licensee's organization;
• train, fine-tune, evaluate, and benchmark Models with the Dataset;
• create Derived Data for the Licensee's internal use; and
• use, sell, and distribute Models and their outputs, including for commercial purposes.

Employees of the Licensee may use the Dataset. So may contractors and cloud providers that work for the Licensee under a written duty of confidentiality, and only for the Licensee's benefit. The Licensee is responsible for what they do with it.

If the Licensee is an individual, only that individual may use the Dataset, and not for the benefit of an employer or a client. An employer or a client needs its own license.

Affiliates of the Licensee need their own license unless oto agrees otherwise in writing.

The Dataset may include Public Sessions. In the full set these are 125 sessions, about 33.5 hours. A copy of a Public Session taken from oto's public releases stays under the Creative Commons license, and nothing in this license limits what that license allows.

oto keeps every right it does not grant here. The license lasts until it ends under section 11.

4. Restrictions

The Licensee must not:

• share, publish, resell, rent, or sublicense the audio, the metadata, or Derived Data that reproduces a substantial part of them;
• identify or try to identify any speaker, contact any speaker, or combine the Dataset with other data to do so;
• create a synthetic voice that imitates an individual speaker, or build a Model designed to reproduce the voice of an individual speaker;
• use the Dataset to recognize or verify the identity of the speakers in it, or to build a collection of their voiceprints;
• try to restore audio that was muted for privacy;
• use the Dataset, or a Model trained on it, to make a dataset of recorded or synthetic conversation that is offered to others; or
• use the Dataset in breach of any law that applies to the Licensee.

Transcripts of the recordings, and labels tied to single sessions, are Derived Data that reproduces the Dataset. The Licensee may make them for its own use and may not share or publish them.

The Licensee may describe the Dataset and report results and statistics across many sessions in papers and product documentation. It may not publish audio from the Dataset.

5. Speaker privacy and removal

Every session was recorded with the consent of both speakers. A speaker may still have a legal right to have a recording removed, and oto may have to remove a session for other legal reasons.

If that happens, oto will send a Removal Notice to the notice email given at checkout. Within 30 days of the notice, the Licensee must:

• delete the listed sessions, and any Derived Data that reproduces them, from every system it controls; and
• stop using the listed sessions for new training or evaluation.

The Licensee does not have to retrain or delete a Model that was trained before the notice, unless a law that applies to the Licensee requires it.

A Removal Notice does not give the Licensee a right to a refund or to replacement sessions.

The Licensee must keep its notice email current by writing to agent@oto.earth, and must confirm deletion in writing if oto asks.

6. Data protection

The recordings contain the voices of real people. oto and the Licensee each decide for themselves how they use that personal data, and each is responsible for following the data protection laws that apply to it.

If a speaker contacts the Licensee about a recording, the Licensee must tell oto at agent@oto.earth within 10 days. The Licensee may answer the speaker as the law requires of it, and otherwise leaves the reply to oto.

Where the law requires a transfer agreement for the Licensee to receive the recordings, such as standard contractual clauses, the parties will sign one.

7. Security and confidentiality

The Licensee must keep the Dataset confidential and protect it with reasonable technical and organizational measures. It must tell oto without undue delay if it learns that someone accessed or shared the Dataset without permission.

8. Ownership

oto and its licensors keep all rights in the Dataset. This license does not transfer ownership.

The Licensee owns its Models and their outputs.

Publications that use the Dataset are asked to cite it as shown on the product page.

9. Warranties

oto warrants that it has the right to grant this license, and that every speaker in the Dataset confirmed that they were an adult and agreed that oto may license the recordings to others for commercial use, including the training of AI models.

Apart from that, the Dataset is provided as is. To the extent the law allows, oto gives no other warranty, express or implied, including warranties of merchantability, fitness for a particular purpose, and non-infringement. Speaker profile fields are self-reported and may be incomplete or wrong.

If someone claims that the Dataset as delivered infringes their rights, oto will send a Removal Notice for the affected sessions. Apart from a claim for breach of the warranty in this section, which is subject to the limits in section 10, this is the Licensee's only remedy for such a claim.

10. Liability

Neither party is liable to the other for indirect or consequential loss, or for lost profit or revenue.

Each party's total liability under this license is limited to the amount the Licensee paid for the order.

These limits do not apply to a breach of section 4, to the Licensee's duties under section 5, to fraud, gross negligence, or wilful misconduct, or to liability that the law does not allow to be limited.

The Licensee will cover oto's losses from third-party claims that arise from the Licensee's breach of section 4. oto will tell the Licensee about such a claim promptly, let the Licensee take part in the defense, and not settle it without asking the Licensee first.

A breach of section 4 can cause harm that money cannot repair. oto may ask a court to stop it without first proving loss.

11. Term and end of the license

The license starts when payment is complete.

oto may end the license by written notice if the Licensee breaches it and does not fix the breach within 30 days of being told. For a breach of section 4, oto may end the license at once.

The license also ends at once if the payment for the order is refunded in full or reversed.

When the license ends, the Licensee must delete the Dataset and any Derived Data that reproduces it within 30 days, and confirm this in writing. Models built in line with this license may still be used.

Sections 4, 5, 6, 8, 9, 10, and 13 continue after the license ends.

12. Compliance

The Licensee must follow the export control and sanctions laws that apply to it. Once a year, if oto asks, the Licensee will confirm in writing that it is following this license.

13. General

This license and the Terms of Sale are the whole agreement about the Dataset. If they conflict, this license applies to the use of the Dataset.

The version of this license shown at checkout applies to the order. oto records the version and the time of acceptance.

The Licensee may not transfer this license without oto's written consent, except to a successor in a merger or a sale of substantially all of its business that accepts this license in writing.

The Terms of Sale say when oto may name the Licensee as a customer and how the Licensee can stop it.

If a court finds part of this license unenforceable, the rest stays in force. A delay in enforcing a right is not a waiver of it.

This license is governed by the laws of the State of California, United States, without regard to its conflict of laws rules. The state and federal courts in San Francisco, California have exclusive jurisdiction.

If the Licensee is an individual buying as a consumer, the consumer protection laws of their country that cannot be waived still apply, and they may bring a claim in the courts where they live.

Notices to oto go to agent@oto.earth. Notices to the Licensee go to the notice email given at checkout.