Description
The dataset includes audio recordings from the University forest of the Philipps University of Marburg (Marburg Open Forest). The Marburg Open Forest is a temperate european forest located about 10km from the city center of Marburg (Lahn), Germany. 48 AudioMoths were distributed throughout the forest for the recordings. Recordings were 48 kHz 16 bit mono. A recording is always one minute long. The recording rhythm was 1 minute recording then one minute pause in which no recording took place. Recordings were made 24 hours a day for around two weeks every month. The files are named according to the recording date using one of the following two naming conventions: Either Year_Month_Day__Hour_Minute (e.g. 2021_10_01__00_00) or Year_Month_Day__Hour_Minute_Second (e.g. 2021_10_31__23_58_03). The audiodata was recorded from April 2020 to October 2022. To comply with privacy regulations, human vocalisations have been detected and removed from the dataset using voice activity detection (VAD). The highly sensitive VAD detector Silero VAD has been applied (https://github.com/snakers4/silero-vad). The entire dataset consists of 14093659 files in WAV format and is around 77 TB large. Due to the size of the complete data set, provision via download is not practical. The complete dataset can be requested by clicking on the file complete_dataset.bit. The full dataset is available upon request from 01. January 2024 onwards.