Turning local soundscapes into useful wildlife evidence.
Project S.W.A.N. uses passive acoustic monitoring to listen for birds and other wildlife sounds, process them with AI-supported identification, and turn the results into public information that can support education, local awareness, habitat improvement and long-term conservation thinking.
The goal is not simply to make a list of birds. The goal is to build a living picture of how wildlife uses local spaces over time: which species are active, when they are most vocal, how activity changes through seasons, and how better habitats can protect birds from avoidable damage.
Method
Passive
Stations listen without handling, attracting, chasing or disturbing wildlife.
Analysis
AI-aided
BirdNET-style acoustic recognition turns short sounds into likely species detections.
Output
Evidence
Species, time, confidence, station context and audio clips become explainable records.
Purpose
Protection
Better evidence helps people notice wildlife before habitats are damaged or forgotten.
What Project S.W.A.N. is doing
Listening stations create a long-term picture of local wildlife activity.
Many birds are easier to hear than see. A robin singing from cover, a wren calling from a hedge, or a blackbird singing at dawn can all be missed during a short human visit. A listening station can keep watch for long periods and collect repeat evidence without needing people to stand nearby.
This matters because wildlife protection often depends on awareness. If people do not know which species are using an area, it is easier for habitats to be overlooked, over-tidied, disturbed or damaged. Acoustic data helps make hidden wildlife activity visible.
The station listens
A microphone records short soundscape sections from a school, garden, woodland edge, village, public site or other listening location.
The sound is analysed
AI-supported acoustic recognition checks the clip and suggests likely species based on patterns in bird calls and songs.
The record is stored
The project stores the likely species, time, date, confidence value, station and, where available, a short audio clip.
People can understand it
Project S.W.A.N. turns technical detection data into station pages, public summaries, classroom dashboards and educational activities.
Why collect this data?
The value is in patterns, not just individual detections.
One detection can tell us that a bird was probably heard at a particular time. Thousands of detections across days, weeks and seasons can show much more: repeated activity, daily rhythms, seasonal changes, dawn chorus strength, habitat differences and possible changes over time.
📍
Local species presence
Repeated detections help build a picture of which birds regularly use a place. This can support local wildlife records and give people a clearer understanding of the species around them.
🌅
Daily activity patterns
Birdsong often changes through the day. Stations can show early morning peaks, quieter periods, evening activity and changes caused by weather, season or disturbance.
🍂
Seasonal change
The same site can sound different in spring, summer, autumn and winter. Long-term data can show when species become more vocal, arrive, leave, breed or become quieter.
🌳
Habitat comparison
A station near mature trees may detect different activity from one beside open playing fields, roads, farmland or water. Comparing sites can help people understand why habitat quality matters.
⚠️
Early warning signs
If a normally active location becomes unusually quiet, or if a species disappears from the acoustic record, it can trigger questions and encourage closer investigation.
🧑🏫
Education and engagement
Children and local communities can explore real wildlife data from places they know, making conservation feel local, practical and alive.
Why the nodes matter
A network of nodes is more powerful than one listening point.
A single station can tell us about one place. A network can show how different habitats behave, how wildlife activity changes between locations, and where local biodiversity may need more attention.
01
Coverage
More nodes mean more listening locations. This helps show whether activity is local to one site or repeated across a wider area.
02
Comparison
A school garden, woodland edge, open field and wetland edge may all produce different soundscapes. Comparing them helps explain why habitats matter.
03
Continuity
Repeated listening creates a long-term record. This is useful because nature changes slowly, and short visits can miss important patterns.
04
Resilience
If one node is quiet, offline or affected by weather, other nodes can still provide useful context across the wider network.
Network effect
The more places we listen, the better the local picture becomes.
Nodes can help schools, communities and future conservation partners ask better questions: which places are rich in bird activity, which habitats are quieter, which species are detected repeatedly, and where could small improvements make a difference?
Helping prevent damage to birds
Data helps people notice birds before habitats are disturbed.
Birds are often affected by damage that looks small at first: hedge cutting at the wrong time, removal of dense cover, loss of berry bushes, disturbance near nesting areas, excessive lighting, heavy tidying of leaf litter, or changes that reduce insect food.
Project S.W.A.N. cannot physically protect a bird by itself. What it can do is provide evidence, visibility and education. When people know that wildlife is using a place, they are more likely to make careful decisions before changing it.
Avoiding accidental disturbance
Repeated bird activity can remind landowners, schools and community groups to think carefully before cutting hedges, removing shrubs, clearing nesting cover or carrying out noisy work during sensitive periods.
Protecting food sources
Many birds rely on insects, worms, seeds and berries. Data can support conversations about leaving wild corners, planting native flowers, keeping berry bushes and reducing unnecessary chemical use.
Noticing quiet places
A quiet soundscape is not automatically bad, but it can be a useful prompt. Is there enough shelter? Are there insects? Is the area too exposed, too noisy, too bright or too tidy for wildlife?
Checking whether improvements help
If a school adds a pond, hedge, bird boxes, long grass or native planting, the station can continue listening afterwards. Over time, the data may help show whether bird activity changes.
🌿
Before damage
Listening data can show that birds are using an area before work takes place. This helps people pause, check and plan more carefully.
🛠️
During change
Where appropriate, monitoring can help compare sound activity before, during and after changes such as habitat improvement or site disturbance.
🐦
After improvement
If habitats are improved, ongoing listening can help communities see whether birds continue to use the site over time.
Data pipeline
From a sound in the environment to a live station record.
Each record on Project S.W.A.N. is part of a chain. The station hears the environment, the software analyses likely species, and the website explains the result in a way that people can understand.
01
Capture
A station records a short section of the soundscape using a microphone.
02
Analyse
AI-supported software checks the sound against known bird vocal patterns.
03
Score
The detection receives a confidence value showing how strongly the model supports it.
04
Store
Time, species, station and audio information are stored for later viewing.
05
Explain
The website turns records into public station pages, maps and classroom tools.
What is collected?
The project focuses on wildlife signals, context and evidence.
The useful part of the data is the environmental signal: what species was likely heard, when it was heard, where the listening station is, how confident the model was, and whether a short audio clip is available.
This makes the data useful for education and conservation without needing intrusive wildlife surveys. It is designed around passive listening rather than physical contact with animals.
Likely species
The bird species suggested by the acoustic model, such as Robin, Blackbird, Wren, Blue Tit, Jackdaw or Wood Pigeon.
Time and date
The exact time of the detection, allowing activity to be compared across hours, days, seasons and years.
Confidence score
A model-generated confidence value. It helps users understand how strong or uncertain the detection may be.
Audio clip
Where available, a short soundscape clip helps people hear the evidence behind the detection.
Station context
Each detection is linked to a station, helping people understand the broad location and habitat context.
Trend potential
Over time, repeated records can support summaries, comparisons, habitat reviews and public education.