Sound Recording and Labelling #
Callan Alexander - PhD candidate at QUT
Recorder settings #
Sample Rates #
- If in doubt, use a higher sample rate.
- You can always downsample later. A higher sample rate will ensure greater compatibililty with more projects.
- Main downside to higher sample rate is storage space, power draw is usually negligible
- 44/48kHz preferable and is becoming industry standard for most terrestrial fauna (apart from bats etc.)
- Ideally only use a lower sample rate if you have very specific requirements, e.g. Acoustic Observatory
Batteries/SD Cards #
- Follow manufacturer guidelines
- For AA - Energiser Pro Max or Eneloop Pro. Disposables unfortunately last longer than rechargable
- Always check and format SD cards prior to deployment.
Deployment Tips #
- Gaia GPS
is widely used by off-track hikers and really good for marking recorder GPS locations even with 0 mobile reception. Make sure to pre-download the map area if you anticpate having no signal. - Taking photograph of each recorder highly reccomended (and keeping them as far away from paths as possible)
- Using Gaia + photograph = no recorders lost yet in hundreds of deployments
- Remember to always check weather and park alerts before AND during deployment - you don’t want to lose equipment to floods and/or fire
Annotation Overview #
- Why Annotate?
- This will be mainly linked to your research question, but a few examples are: call description
, getting species list for an area , and building and improving models
- This will be mainly linked to your research question, but a few examples are: call description
- Important to consider how your annotations will be used prior to commencing
- How accurate do your measurements need to be? For some questions you may need to tag multiple notes with high accuracy, but for other questions a loose box can be sufficient
- Multi-species vs single-species
- Whole vocalisation or separate notes?
- This will be covered in more detail in the practical
Using Raven #
- Why Raven?
- Raven Pro vs Lite
- Biggest disadvantage of Lite is only one annotation column and less measurements
- Be careful of focus settings, many cases of people tagging vocalisations that are out of focus
- Hanning Window is most commonly used
- Focus can vary depending on the recorder but for an hour long file, 2132 is often a good starting point
Manual Scanning for Vocalisations of Interest #
- Can drag multiple hours of data into Raven and page depending on computer specs, or just work with an hour at a time.
- Do not tag when you are working with multiple files, Raven does not separate the annotations. Rather find a file of interest and tag it seperately
- WarbleR - Spectrogram to Pdf Tool useful in some scenarios
Packages to use with Raven #
- WarbleR
- RRaven
- Tidyverse
- AutoHotkey
E.g. Generate pdf spectrogram of multiple files for rapid scanning Group multiple raven files into one csv file, generate spectrograms for all detections Will cover some of this in the practical!
Useful Autohotkey script #
This scripts allows you to use your arrow keys to move left and right.
Note: You will need to download Autohotkey
#NoEnv ; Recommended for performance and compatibility with future AutoHotkey releases. ; #Warn ; Enable warnings to assist with
detecting common errors. SendMode Input ; Recommended for new scripts
due to its superior speed and reliability. SetWorkingDir %A_ScriptDir%
; Ensures a consistent starting directory.
#IfWinActive ahk_class SunAwtFrame Right:: Click, 1824 790 Click, 1824 790 Return
Left:: Click, 235 791 Click, 235 791 Return
General Useful Resources #
- Apps: Stewart Sounds App
, Morcome & Stewart App , Frog ID - Web Resources: Xeno Canto
, eBird Australian Bird Calls Facebook Group Bioacoustics Stack Exchange - A list of free and paid software available for bioacoustic analyses: https://github.com/rhine3/bioacoustics-software
- Izotope RX8 - batch noise removal, pretty spectrograms, fine-scale sound editing
- Audacity - free noise removal + sound editing
Go to the practical.