Abstract: A method and apparatus is described for generating and recording acoustic piano performances automatically and at scale. This technique has generated nearly 500 hours of studio piano recordings, including a complete re-recording of the MAESTRO solo piano dataset. A software utility has been implemented to simultaneously manage MIDI playback on a Yamaha Diskalvier and a synchronized audio capture session. As part of this late-breaking demo, we release the re-recorded studio audio for the MAESTRO Studio dataset and the Python software package piano-capture used for automating data collection. As well as adding to the pool of existing training data for MIR tasks such as transcription and audio-MIDI alignment, this work facilitates the creation of further datasets, for example in other musical styles.