Archaeological field documentation is still largely based on handwritten context sheets. While this method is well established, it can be time-consuming and difficult to use in typical excavation conditions, such as wet or muddy environments. Handwritten recording can also present accessibility challenges for some archaeologists, for example for those with dyslexia. Recent advances in automatic speech recognition offer new possibilities for supporting field documentation through voice input. This poster and demo present a sound-based approach to recording archaeological context sheets using speech-to-text technologies. The work explores the use of voice recordings as an additional data type within archaeological documentation workflows. The main aim is to support data entry for context sheets by allowing archaeologists to dictate information instead of writing it by hand. The focus is on free-text fields such as context descriptions and interpretations, which are often the most time-consuming to complete and benefit from more natural and detailed expression. Speech-based input may also encourage more complete descriptions by reducing the need for shorthand or abbreviated notes. By using speech, archaeologists can potentially document their observations more efficiently while remaining engaged with excavation activities.

A browser-based web application was developed to support this approach. The application allows users to create and complete digital context sheets using spoken input through their phones or laptops (Figure 1). The system is designed for use in the field and runs entirely in the web browser, without requiring specialised hardware or software installation. The resulting transcriptions can be reviewed and edited through a simple dashboard interface, which also provides options to manage, update, and export completed context sheets (Figure 2). An export function generates data in CSV format, enabling integration with existing archiving systems, which supports the reuse of data.

To support development and testing, a small speech dataset was created using readings based on existing archaeological context sheet descriptions. This material was used to evaluate automatic speech recognition performance in an archaeological domain and to inform interface design. Initial field trials of the web app were carried out during the 2025 excavation season at the Toumba Serron site in Greece. These trials highlighted the importance of low-latency speech recognition for practical field use and led to the adoption of a faster speech-to-text model, improving responsiveness and usability.

The demo session will showcase the web application and its workflow, including live voice-based entry of context sheet data, review and correction of transcriptions, and export of results for reuse in downstream systems. The poster will outline the motivation, technical approach, and lessons learned from early testing. Together, they illustrate how sound-based documentation can complement existing archaeological recording practices and support more flexible, accessible, and efficient field workflows at the intersection of cultural heritage, digital humanities, and language technologies.

https://ibb.co/R4BLZnQr

Figure 1: Interface for adding a new context sheet using voice commands

https://ibb.co/4Z6nWJQF

Figure 2: User dashboard listing completed context forms, with options to edit, delete, and export