> For the complete documentation index, see [llms.txt](https://twbvoiceplaybook.clearglobal.org/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://twbvoiceplaybook.clearglobal.org/8.-access-to-datasets-and-models/8.1-sharing-data-on-hugging-face-and-platform-access.md). # 8.1 Sharing data on Hugging Face and platform access ### What is Hugging Face? Hugging Face is a popular online platform where researchers and developers share AI models[^1], datasets[^2], and applications. It’s like a library for [artificial intelligence](#user-content-fn-3)[^3] resources that many people can use. The platform makes it easy to find, download, and use machine learning models and datasets. These are created by organizations and individuals worldwide. ### Accessing TWB Voice datasets You can find our datasets, models and demos on the CLEAR Global organization page:[ huggingface.co/CLEAR-Global](https://huggingface.co/CLEAR-Global) We publish each dataset with lots of metadata[^4], including number of people taking part, total hours of recordings, and license information. We usually release our datasets under [Creative Commons](#user-content-fn-5)[^5] or OpenRAIL[^6] licenses. These allow broad use, but make sure that people are credited properly, and provide guidelines on use. We publish models with information on their architecture and on their evaluation on a [test partition](#user-content-fn-7)[^7] of the data we publish. You can download the data through the web interface. Or you can use the Hugging Face datasets Python module to access them programmatically[^8] for your projects. ### Important limitations and things to consider The models we share on Hugging Face are mostly research demonstrations and evaluation tools. They have been tested on specific datasets and may not work well in all real-world settings. These models are not guaranteed for production use. You would need to test them thoroughly before using them in any active humanitarian technology system. When using our models, consider the following: * Performance may vary a lot with different accents, recording conditions, or speaking styles. * Models are trained on limited data so they may not represent the full range of speakers in a language community. * For any practical application, they will need regular testing and adjustment. We suggest that users see these models as starting points for further development. They are not complete solutions that could be put to immediate use in critical humanitarian settings. ### Gated datasets Some [TWB Voice](#user-content-fn-9)[^9] datasets are configured as "gated" on Hugging Face. This means you need to log in to the platform and request access before you can download them. We choose to gate some datasets so we can keep track of who is using our data. We can also make sure they use it responsibly. This helps us understand the impact of our work. We can also be accountable to the communities who record their voices for us. When you access gated datasets, you'll need to give some basic information about your intended use. Access is then usually granted automatically.
[^1]: **Model:** A computer-based system that has made use of data to learn patterns. It can make predictions or generate language. In speech technology, models use voice data to learn to recognize speech (converting audio to text) or to create speech (converting text to audio). [^2]: **Dataset:** A collection of information that has been organized for use. A **voice dataset** is a collection of voice recordings (paired with transcription) with additional information (metadata) such as gender, age of the person recording to give more information on how the data set is constructed and to avoid bias. It is for use in research and for training or improving voice models. [^3]: **Artificial Intelligence (AI):** Artificial Intelligence (AI) refers to computer systems that can do tasks that normally require the intelligence of a human. AI allows computers to do things that usually only humans can do. It’s like having a smart friend who can learn, make decisions, and solve problems. But you don’t need to program them for each new task. [^4]: **Metadata:** Extra information that gives some background to a dataset or parts of a dataset. Examples are demographics of the speakers (age, gender, accent), recording conditions (microphone type, level of background noise), or language-related details (dialect, speed of speaking, emotional tone). [^5]: **Creative Commons (CC) licenses:** A set of standard copyright licenses that let the creator of a piece of work choose which rights they want to keep and which they will give up. [^6]: **Open and Responsible AI License (OpenRAIL):** A licensing framework that allows wide use but also provides responsible guidelines for the use of AI. [^7]: **Test partition:** A separate portion of data that is used to find out how well a model is working. [^8]: **Programmatically:** Accessing data through code and programming interfaces rather than by manually downloading it. [^9]: **TWB Voice:** A platform for collecting voice data. It was developed by CLEAR Global, who also own it. Users can make voice recordings to help with active data collection projects in TWB Voice by [signing up to the TWB Community](https://translatorswithoutborders.org/join-the-twb-community/). The main goal of TWB Voice is to help to develop voice technology for speakers of marginalized languages. For example, by creating the voice datasets that are needed to build language models for TTS and ASR.