docs/data-sets.md

# Audio datasets

> We have automatic queries running that will keep updating and uploading more unlabelled data over the next weeks. So this dataset is still growing. If not stated differently, all data was sampled at a frequency of 48KHz. The document sampling_sites has more information on local observations.

This audio data was collected by mentors [email protected], David Dao, [email protected] and @Marina Gatto

The data was collected through two sensors:

- Audiomoths
- SongMeters

**_likely_bird_songs_in_Ingles_**
This file contains a list of likely bird songs to occur in the area of Ingles (Northern Rio Negro) and was curated by our domain experts. It is not exclusive.

**_Ingles_**
All audiomoth collections of soundscapes in Ingles.
This contains the primary forest (labeled as primary) as well as in an Inn (with direct human pressure).
The audiomoth in the primary forest was recording up to a maximum frequency of 192Khz (creating a much larger dataset) with the hope of capturing ultrasonic soundscapes (bats)

**_SongMeters_Ingles_Primary_**
All song meter recordings of the soundscapes of the primary forest of Ingles.

**_SongMeter_Clusters_**
SongMeter Clusters are unsupervised detected clusters (run by Mentor [email protected]). They contain frequently occurring patterns detected in the SongMeter recordings in the frequency range of 2-5KHz.

**_Inhaa-Be_**
3K min. All audiomoth collections of soundscapes in a forest close to Inhaa-Be.
Inhaa-Be is an Indigenous Village, with a protected forest.

**_ParqueDasTribos_**
2K min. All audiomoth collections of soundscapes of Parque Das Tribos.
Parque das Tribos is an Indigenous Urban Village at the outskirts of Manaus. Most recordings are close to roads and human pressure.

**_Xingu_**
14K min. All this data was collected in a previous trip in Para State (a different state than Manaus) in October to November 2023. It is deployed in a protected territory far away from any human pressure. None of the provided global layers overlap with this area but can be requested if needed.

**_XenoCanto/Amazonas_**
All XenoCanto songs that are originating from “Amazonas” (can include Colombia etc). This folder is split between frogs and birds.

**_XenoCanto/Greater Manaus_**
A smaller subset of XenoCanto songs, containing all observations of the Upper Rio Negro area (overlapping with Manaus, Inhaa-Be, Ingles and Parque Das Tribos)

## primary 1,2 (including ultrasonic)

- location:
  - In the primary forest
- files

  - WAV
  - each 44 MB
  - Sample rate (Hz) : 384000
  - Sleep duration (s) : 240
  - Recording duration (s) : 60

- links:

  - primary 1: https://drive.google.com/open?id=1VCP9VDDtm6-u0lPsm6Pi_A0nBNvQjPv9&usp=drive_copy

  - primary 2: https://drive.google.com/open?id=1sIeJ_1GBHuU7GO5CU6WVXZycU7I5YA_B&usp=drive_copy

## SoundMeters_Ingles_Primary

- location:
  - In the primary forest
- files
  - WAV
  - 10.6 MB
  - Recording duration (s) : 60
- link
  - https://drive.google.com/open?id=1eu0P_PrTjgVhNVZK8xycYlDd5AnJ9WdN&usp=drive_copy

## inhaa-Be Audiomoth 1,2

- location:

  - In a protected forest of Inhaa-Be

- files

  - WAVE
  - each 5.5 MB
  - Sample rate (Hz) : 48000
  - Sleep duration (s) : 240
  - Recording duration (s) : 60

- links
  - https://drive.google.com/open?id=1wUj3rxruAqInWgJzumFTYaetVbQ8YfW2&usp=drive_copy
  - https://drive.google.com/open?id=1raI3UcCUWKg49tIE6L9LdfRlE557mlNp&usp=drive_copy

## inn 2, 3

- location:

  - inn 2: At the Garden of the Inn (human settlement)
  - inn 3: At the Inn, close to the River (human settlement)

- files

  - WAV
  - each 5 MB
  - Sample rate (Hz) : 48000
  - Sleep duration (s) : 5
  - Recording duration (s) : 55

- links

  - inn2: https://drive.google.com/open?id=18fy059Ypaq7kYnjZm4TNLoqkfmOHhaIE&usp=drive_copy
  - inn3: https://drive.google.com/open?id=1wUj3rxruAqInWgJzumFTYaetVbQ8YfW2&usp=drive_copy

## parque das Tribos

- location:
  - Close to a road, at the outskirts of Manaus (human settlement)

## landing - (includig ultrasonic)

- location:
  - In the secondary forest / recently deforested and close to a flooded area