The status of forensic speech recordings among existing data protection guidance is not clear. The inherent nature of voice and the way in which forensic speech casework is currently allocated mean that there are additional barriers to incorporating real casework data into research activities. The key objective of this work is to explore data protection solutions that could enable the forensic speech science community to responsibly use real casework data for research and development purposes. While reviewing relevant guidance and rulings, issues such as proportionality, opportunism and data minimisation are addressed, as well as where voice sits in relation to the definition of “biometric data”. This paper ultimately places forensic speech recordings in the data protection context to illuminate the specific issues that arise for this data type.