Dendrophone is a site-specific sound installation by Peter Batchelor located in Alice Holt Forest, Surrey, UK, that transforms local environmental data into immersive sound textures. These recordings are made directly from the installation’s multichannel output and capture both its generative soundscape and the ambient natural environment. The sounds respond in real-time to three key ecological variables:
• Humidity – represented sonically by dry, crackling textures or damp, flowing ones depending on forest moisture levels.
• Sunlight energy – conveyed through shifting hissing sounds, juddery when photosynthetic activity is high, smoother when it's low.
• Carbon dioxide levels – expressed through breathing-like sounds: long and steady when uptake is high, shorter and erratic when it's reduced.
The installation runs on a DIY multichannel system based on Raspberry Pi Zero microcomputers and low-energy amplifiers. The Raspberry Pis run custom Pure Data patches which process environmental data drawn from sensors in the forest and generate the audio—pre-composed multichannel textures which are triggered and modulated live in response to data. The result is an ever-changing soundscape that reveals the hidden rhythms of the forest.
The surrounding habitat features a variety of native broadleaf species including oak, sweet chestnut, birch, and willow. Wildlife in the area includes roe and muntjac deer, bats, and a variety of birds such as chiffchaff, robin, wren, coal tit, and tawny owl.
The installation runs daily from 11am to 5pm (local solar time), powered entirely by solar panels with lead-acid battery storage. The system is designed to be self-sustaining, though may occasionally go offline due to low sunlight, particularly in winter months.
This automatic recording is part of an ongoing series captured directly from the Dendrophone system and uploaded regularly. The recordings are captured with a DIY Raspberry Pi audio streamer designed by Luigi Marino. This solar-powered system is typically stable, but it may go offline due to extreme weather factors. From 15 Feb 2025 until 20 May 2025, a combination of 4-6 recordings were done using MEMs microphones, capturing both solar times (sunrise, solar noon, sunset, and at the midpoint between sunset and sunrise) and installation times (e.g. mid-morning and mid-afternoon). Since 20 May 2025, the quality of the recordings has improved considerably because of changes in the equipment, including switching to Rode LavalierGO mics with Rode AI-Micro audio interface and software updates. The frequency of the recordings has also changed to be captured 10 times a day according to solar time (sunrise, solar noon, sunset, and at the midpoint between sunset and sunrise) and 11am to 5pm (one recording per hour).
This dataset is funded by the AHRC Sensing the Forest project (AH/X011585/2). More info at: https://sensingtheforest.github.io
Type
Wave (.wav)
Duration
5:00.000
File size
50.5 MB
Sample rate
44100.0 Hz
Bit depth
16 bit
Channels
Stereo
Comments
