Back to billewood.com

BloomingSongs

Mapping when birds sing across America

An open-source project visualizing the seasonal rhythms of bird vocalizations, revealing when and where different species fill the air with song.

The Question We're Answering

When exactly do birds sing? Not just "spring" — but which week? Does it differ between California and New York? And why doesn't this data already exist?

Birders know that bird song peaks during breeding season. But surprisingly, there's no easily accessible public dataset showing when birds vocalize throughout the year, broken down by species and location.

Existing resources like eBird show when birds are observed (seen or heard), but don't distinguish between silent winter flocks and the dawn chorus of spring. BloomingSongs fills this gap by analyzing audio recording data to reveal true vocal phenology — the science of seasonal timing.

46,000+
Audio recordings analyzed
573
Bird species
8
US states covered
44
Years of data (1979-2023)

Why This Matters

🎯

Better Birding

Know the exact weeks when your target species is most vocal in your area, maximizing your chances of hearing (and finding) it.

🔬

Climate Research

Track shifts in singing phenology over time. Are birds singing earlier than they did decades ago? The data can tell us.

🏡

Acoustic Ecology

Understanding when birds vocalize helps urban planners reduce noise pollution during critical breeding communication periods.

📱

App Development

Build smarter bird ID apps that account for what species are actually singing at a given time and place, not just present.

A Tale of Two Coasts

One of the first questions we explored: do birds sing at the same time across the country? We compared California and New York for five common species.

Seasonal singing patterns: California vs New York

Figure 1: Weekly singing activity for five species, comparing California (left) and New York (right). Each colored line represents a different year (2019-2023), with the black line showing the multi-year average and gray shading indicating confidence intervals.

Key Finding: Birds consistently sing later in California than in New York — often by 3-7 weeks. The mild California climate extends the breeding season, while New York's harsh winters compress vocal activity into a narrow spring window.

Species Spotlight: Song Sparrow

The Song Sparrow perfectly illustrates this east-west divide. In New York, singing peaks sharply in April when birds rush to establish territories after the spring thaw. In California, the peak comes in May, and there's notable vocal activity even in winter months.

Song Sparrow: California vs New York

Figure 2: Song Sparrow singing patterns. Note California's winter singing (December-February) which is almost entirely absent in New York.

The Exception: Anna's Hummingbird

Not all birds follow the "spring singing" pattern. Anna's Hummingbird, a Pacific Coast resident, has one of the most unusual breeding schedules in North America — and our data captures it beautifully.

Anna's Hummingbird singing across Pacific Coast states

Figure 3: Anna's Hummingbird vocal activity across Washington, Oregon, and California. Unlike most birds, this species shows winter peaks.

Key Finding: Anna's Hummingbirds in California sing most actively in December and February — they're among the earliest nesting birds in North America, often with eggs in the nest by January. The summer trough corresponds to their post-breeding molt when males sing less.

The data also reveals a clear latitudinal gradient: California dominates with nearly 200 recordings, while Oregon has just 14. This matches the species' range — they're abundant in California but scarce in the Pacific Northwest.

What We're Really Measuring

Any analysis of citizen science data must grapple with a fundamental question: are we measuring bird activity, or human activity? The answer is both.

Total bird recordings by month

Figure 4: Total bird audio recordings by month. This represents recorder effort — when people are out making recordings — not necessarily when birds are most active.

In New York, recording effort is highly seasonal: almost 4,000 recordings in May, but fewer than 200 in January. California is more even year-round. This means raw counts can be misleading — a species might appear to sing less in winter simply because fewer people are out recording.

Monthly recording counts with effort correction

Figure 5: Species-specific recordings (blue bars) with effort-corrected frequency (red line = percentage of all recordings). The red line helps reveal true seasonal patterns independent of recorder effort.

Methodology Note: Our analysis uses both raw counts and effort-corrected frequencies. For cross-region comparisons, we apply smoothing and confidence intervals to account for year-to-year variation and sampling uncertainty.

Open Data, Open Science

BloomingSongs is built entirely on publicly available data. Our primary source is the iNatSounds dataset — a collection of 230,000+ wildlife sound recordings from iNaturalist, filtered to ~45,000 US bird recordings.

Unlike eBird observations (which include birds seen silently), every record in our dataset is an audio recording. This makes it a genuine proxy for vocal activity, not just presence.

iNatSounds

45,435 recordings
1979-2023
Audio only

eBird API

1,277 records
Recent observations
Presence data

eBird Breeding Codes

22 records
"Singing" codes (S, S7)
Explicitly vocal

All code, data processing scripts, and documentation are open source. We believe in transparent, reproducible science.

What's Next

BloomingSongs is an evolving project. Here's what we're working toward:

Interested in contributing? The project is open source and welcomes collaborators with skills in data science, ornithology, or web development.