User guide
In the dashboard, administrators can create projects and edit their picture, URL, short and extended descriptions. The extended description appears only on the project's page. Projects can be inactivated to remove them from the default, public overview page.
Administrators and project managers can create users and edit their ORCID, email, role (user/administrator), password and status (active/inactive) as well as the color of their tags. The privileges of users for single collections can be set up here (list icon button), and users can be given privileges for managing projects to become project managers. The recordings and collections created/uploaded by users will receive an ORCID badge (if specified) linking to the public profile of the corresponding user. Users can be inactivated to restrict their access to the website. No deletion is currently possible.
Administrators and project managers can create collections inside their managed projects and edit their fields (DOI, sphere, description, default view). Collections and/or their tags within can be public (accessible to non-logged in visitors) or closed (accessible to specific users). User access to collections and their tags is determined by the user privileges (see "Users"). The collection table data can be downloaded as a CSV, and collections can be deleted with all their contents (recordings, tags) after a security prompt.
Three different layouts are available to view recordings within collections. Site maps displaying recording numbers are visible in every view. Recordings can be opened in the player by clicking their thumbnail or placeholder.
In gallery view, recordings are arranged in a grid, displaying their sound name and the label of the logged-in user. It is possible to search for recordings using the text box, which will search across all fields:
In list view, recordings can be played back directly and some of their meta-data are displayed. It is possible to search for recordings using the text box, which will search across all fields:
In timeline view, recordings are arranged on a timeline (on X axis) and grouped by sites (Y axis). The timeline can be zoomed and panned:
Note: Currently, it can take long (>1 minute) to load collections with several thousands of recordings, even with a fast internet connection.
Administrators and project managers can upload recordings ("Upload files" button), or meta-recordings ("Meta-data" button: allows to upload a recordings CSV list) to specific collections. The date and time of a recording can be automatically extracted from the filename when in the format YYYYMMDD_HHMMSS. Prefixes for the sound names can be given. Other mandatory fields are: recorder, microphone, gain. Optional fields are: site, DOI, Creative Commons licence, recording medium and type.
The meta-recordings or recordings are inserted into the database (for recordings, the MP3 preview and PNG spectrogram are created then) after clicking the "Insert" button. Recordings will appear in the corresponding collection after some processing time elapsed.
The progress of the recording insertion can be viewed in the queue tab of the dashboard. Jobs are tagged as "failed" when warnings occur and these are logged in the corresponding column:
After upload and databse insertion, all recording fields can be edited a posteriori. Multiple recordings can be selected for viewing, deletion, or analysis. Analysis options include acoustic indices (multiple selectable alpha acoustic indices from scikit-maad) and deep learning AI models (BirdNET Analyzer v2.4, batdetect2 v1.0.6). Single recordings can be exported.
Recordings can be opened in the spectrogram player. Recordings can be played back, navigated, labeled, annotated, downloaded and analysed. Buttons with enabled functions are shown with green shading.
The spectrogram is the most prominent part of the player. It visualises sound like a music score: time on the X axis, frequency on the Y axis, and amplitude as a color code (black: silent, red: loud). The current default setting is a window size of 1024 for the FFT-derived spectrogram but this can be changed globally (in the user's admin interface) or per recording (using the utilities menu). If your recording has two audio channels, clicking on the L/R letters left of the main spectrogram switches between the two channels. The current spectrogram frame, if it only shows a part of the recording, can be shifted with the left and right arrow buttons to the left of the main spectrogram. Any selection of the spectrogram can be zoomed into by clicking the zoom button (magnifying glass icon). By default, the filtering option is enabled (filter icon), which will filter out frequencies outside of the selection (what you see is what you hear). You can return to the overview by clicking on the spectrogram thumbnail to the bottom left of the main spectrogram.
You can click the playback button to listen to the audio. For listening to ultrasound calls, the playback speed slider can be adjusted to artificially make them audible. The red cursor shows the position in the recording and can be dragged to play back from any part. The playback time of every user for every recording is logged to be able to derive the sampling/listening intensity for each recording. Continuous playback can be enabled by clicking the downward black arrow next to the playback button, which will shift the spectrogram frame automatically to enable playback until the end of the recording. By clicking on the reading mode button to the left of the spectrogram (CD icon), several pre-defined and also custom visualisation densities can be selected to standardise spectrogram screening. Playback will be started from the start of the recording, over all frequencies, with continuous playback mode enabled.
Any part of the spectrogram can be selected. The coordinates of the selection are shown inside the text boxes below the spectrogram. These values can be copied (delimited by tabs) with the clipboard button next to them and inserted directly into any spreadsheet for further analysis. The frequency at which the highest sound level is detected (FMaxE in bat call analysis) can also be copied to the clipboard with the peaks button. Several alpha acoustic indices (courtesy of scikit-maad) can be selected to analyse any recording or recording part. The entire recording can be run through BirdNET or batdetect2 CNNs to automatically detect and identify bird and UK bat vocalisations. Index analysis results can be saved, accessed, and exported from the dashboard's "Index logs" tab. CNN results are automatically saved as annotations. The progress of the Acoustic index analysis and CNN execition can be viewed in the queue tab of the dashboard.
Tags can be created with the tag button (see dedicated section below) or automatically generated by CNNs.
In the player, tags can be created and appear as boxes overlaid on the spectrogram; they can also be created by executing CNNs (see above) that identify sound sources and store the result as new tags.
Tags can be hidden and displayed using the hide tags button (eye icon) left of the main spectrogram. Only the user's own tags can be seen, but administrators or project managers can give viewing privileges to see other user's tags too.
Three actions appear after clicking tags:
- edit tag: opens the tag window
- zoom tag: zooms into the audio within the tag
- estimate call distance: starts the dedicated call estimation playback, prompts for a distance estimate
Any spectrogram selection can be used to create a tag by clicking on the tag button (tag icon) to the left of the main spectrogram. This, like editing tags, will bring up the following window:
In the "phony" box (soon to be renamed), any soundscape component (bio-, anthropo-, geo-, unknown) can be selected. As there is no standard classification for sound types yet, sound type input is optional and can be customised. Phony and sound type selections are remembered for the next tag. The species box only appears for biophony tags. You can start typing in the English or Latin name of the animal and choose the corresponding matches from the list of animals that you have uploaded in your ecoSound-web installation. Xeno-Canto and Google images links can be clicked to ascertain identities. Our live instance includes bats, birds, and mammals from the IUCN Red List taxonomy. If you are unsure of the ID, check the box “uncertain”.
The frequency and time boxes are automatically filled based on your selection, but the values can be edited manually too. During tag creation, the call distance box is greyed out and the value only appears after using the dedicated call estimation function to estimate distances after a standardised playback (see below). The number of individuals is 1 by default but can be changed to any non-null integer value. Comments can be input. The tag window shows the creator and creation time of the tag. A shareable link can be created for each tag by clicking the green sharing icon.
If the tag contains a reference call or sound, the corresponding box can be checked. In the future, we aim to implement search queries for assembling reference collections dynamically, based on all recordings that the logged-in user has access to.
Animal call types are often described with an “Onomatopoeia”, a word that resembles the sound itself (e.g., like “cuckoo”, “squeak”, ”hiss”). This is subjective and new onomatopoeia tend to created for already described sounds, so we prefer the usage of call types describing the behavior or ecology of the species. Call type categories differ among taxa.
Available call types for primates:
| Call type | Description |
|---|---|
| Advertisement – mating | Emitted during mating interactions (should be temporarily close to a mating event) |
| Advertisement – territory | Emitted for territorial defense and advertisement |
| Affiliative | Emitted during affiliative interaction |
| Agonistic | Emitted during agonistic interaction |
| Alarm | Emitted in alarm context (in response to a predator or disturbance) |
| Begging | Emitted by an infant towards its mother for food or care |
| Contact | Emitted between group members to establish contact |
| Feeding | Emitted in the context of feeding |
| Foraging | Emitted during foraging |
| Searching | Emitted during searching (other than foraging) |
| Social | Emitted in the context of a social interaction (e.g. if affiliative or agonsistic is not known) |
| Song | Call series of repetitive vocal elements. Often exchanged between individuals, and neighboring groups. (In practice, this might often be a duet and involve two or more individuals.) |
Available call types for birds:
| Call type | Description |
|---|---|
| Song | Emitted for territorial defense or mate advertisement |
| Call – begging | Emitted by chicks towards parents |
| Call – contact | Emitted to establish location of members of flock |
| Call – flight | Emitted in flight |
| Call – unspecific | Unknown function |
| Alarm | Emitted as response to predator or other disturbance |
| Non-vocal | For instance drumming of woodpeckers |
Available call types for bats:
| Call type | Description |
|---|---|
| Searching | Emitted during echolocation |
| Feeding | Feeding buzz (for catching prey) |
| Social | Emitted in social context, to communicate with other individuals |
Available call types for frogs:
| Call type | Description |
|---|---|
| Defense – non reproductive | Emitted for defense of diurnal retreats not used for reproduction |
| Discouragement | Discouraging takeover attempts by other males during amplexus |
| Acquisition/defense - reproductive | Acquisition or defense of reproductive territories |
| Advertisement | Advertisement towards males |
| Courtship | Emitted during courtship |
Call distance must be filled only after hearing an unfiltered (frequency-wise) version of the call. Otherwise, filtered audio sounds too unnatural so that the distance estimation could be biased. Distance estimation works only by clicking the distance estimation button (horn icon) after the tag is created. This zooms in to the first 30 s of the tag and plays it back over the entire frequency range. In parallel, users should listen to the recording's corresponding sound transmission recording to gauge their own hearing to the location's and microphones' specific sound transmission. References: Darras et al. 2018 Darras et al. 2016
Note that for bats, the "distance not estimable" checkbox should be checked. As long as tags have no distance data or not explicitly not estimable distance, they will stand our with an orange shading.
If the user has reviewing privileges (or is an administrator or project manager), the right pane of the tag window will appear, which allows to confirm, revise, or mark tags as uncertain or for deletion. The user and action are logged in a list. Only one review is allowed per user and tag. Tags that have not yet been reviewed appear with dashed borders.
Tags data can be edited in the corresponding tab of the dashboard. Multiple tags can be selected for opening the corresponding recording sections, and for deletion. All tags data can be downloaded as a CSV.
