Terms of service

These terms cover two things: your use of this website, and the basis on which we source and license training data. Where a signed agreement exists between us and you, that agreement governs the project and takes precedence over anything on this page.

Using this website

You may read, link to and quote from this site. You may not scrape it at a volume that degrades service for others, republish substantial portions as your own content, or present our material in a way that misrepresents its source.

The content here is provided for information. Descriptions of languages, categories and project process are accurate to the best of our knowledge, but they are not a specification and they are not a warranty that a particular dataset can be produced at a particular price or date. Only a written quote does that.

Enquiries and quotes

Submitting the request form does not create a contract. It starts a conversation. A quote we issue is valid for the period stated in it, and is based on the specification you gave us. If the specification changes, the price and timeline are re-quoted.

We may decline an enquiry. We decline projects involving recorded telephone calls, medical or clinical data, children's voice data without documented guardian consent, and data whose provenance cannot be established.

How a project is agreed

A project begins when a written agreement is in place covering scope, specification, quality criteria, delivery format, schedule, price and re-work terms. The specification is the reference document for quality. Where the delivered data is measured against it, the measurement method is agreed before production rather than after.

Pilot batches are reviewed and either approved or rejected within the period stated in the agreement. Approval of a pilot batch fixes the quality bar for the full run.

Licensing of delivered data

Unless a signed agreement says otherwise, delivered data is licensed rather than sold outright. The license states what the data may be used for, whether the resulting model may be distributed, whether the data may be redistributed, and the territory and duration. Rights we do not explicitly grant are not granted.

You are responsible for confirming that your intended use fits the license you have, and for complying with the law that applies to you. We document the chain of consent and provenance accurately; we do not provide legal advice, and nothing on this site is legal advice.

Redistributing delivered data to a third party, or using it to train a model that will be released under a license that permits redistribution, requires a license that explicitly allows it. Ask before, not after.

Deliverables and documentation

Every delivery includes the data and the documentation described on the compliance page: consent records, collection methodology, the annotation guideline, per-file recording conditions, and the quality report. If any of these is missing from a delivery, tell us and we will supply it.

Confidentiality

Specifications, pricing and delivery schedules are confidential between us. We do not publish who our clients are or what they commissioned. We ask the same of you regarding our sourcing network.

Liability

To the extent permitted by law, our liability in connection with a project is limited to the fees paid for that project. We are not liable for indirect or consequential loss, including lost model performance, lost revenue or lost opportunity.

This limitation does not apply where it would be unlawful, and it does not reduce obligations that a signed agreement states explicitly.

Intellectual property

The text, layout and code of this website belong to us. The datasets we deliver are licensed under the terms of the applicable agreement, not transferred by implication from anything on this site.

Changes to these terms

We may update these terms. The date at the top of this page reflects the current version. Projects already under a signed agreement are governed by that agreement, not by later changes to this page.

Contact

Questions about these terms should go to hello@linguacorpus.com.