xcapi: Xeno-canto API

A Python package for downloading animal sound recordings from the Xeno-canto API. Works with birds, grasshoppers, bats, frogs, and more.

No programming knowledge required — xcapi can be used entirely from the command line.

Features

• Comprehensive search: Support for all Xeno-canto API v3 search tags:
  • Taxonomic: genus, species, subspecies, family, group, English name
  • Geographic: country, location, area, bounding box, latitude, longitude, altitude
  • Quality: rating, sound type, sex, life stage, recording method
  • Temporal: year, month, upload date, time of day
  • Metadata: recordist, length, license, equipment (device, microphone), sample rate, temperature, and more
• Operator and range support: Use comparisons like <, >, and ranges (e.g. 10-20) across supported fields.
• Smart incremental downloads: xcapi remembers what you have already downloaded and only fetches new recordings on repeat runs — no duplicate downloads, no wasted bandwidth.
• Preview before downloading: Use --metadata_only to see what a query would return and how many new recordings would be added, without downloading any audio.
• Two easy ways to use it: Run searches directly from the command line (CLI) without writing any code, or use it inside Python scripts with a simple, chainable query builder.

Installation

pip install xenocanto-api

API Key Setup

Starting October 10, 2025, a Xeno-canto API key is required to download recordings:

1. Register at https://xeno-canto.org
2. Verify your email
3. Get your key from https://xeno-canto.org/account

Output Files

xcapi creates and manages the following files inside your --output_dir:

File	Description
metadata.csv	Metadata for all downloaded recordings. Grows with each download run. Never overwritten unless --redownload is used.
xcapi_runs.json	Internal log of downloaded recording IDs, organised by timestamp. Used to skip already-downloaded recordings on future runs. As IDs are stored by download date, this file also serves as a download history that can be useful for tracking dataset growth over time or for reproducibility purposes.
metadata_only.csv	Created by --metadata_only runs. Contains only recordings not yet downloaded — a preview of what a real download would add. Overwritten fresh on each --metadata_only run.

Audio files are saved into per-species subfolders inside output_dir, e.g. output_dir/Turdus_merula/.

Note: metadata.csv and xcapi_runs.json are only updated by real download runs. Running --metadata_only never modifies them.

Usage

Command Line (CLI)

API key setup

Option 1: Environment variable (recommended)

export XENO_CANTO_API_KEY="your-api-key"

Option 2: Pass it directly each time

xcapi --api_key "your-api-key" --grp birds --cnt Spain

Show all available options:

xcapi --help

---

Downloading recordings

Download all grasshopper recordings from Europe:

xcapi --grp grasshoppers --area europe --output_dir ./data

Download high-quality bird songs from France:

xcapi --grp birds --cnt France --type song --q A --output_dir ./data

Download grasshopper recordings filtered by recordist, altitude, and year:

xcapi --grp grasshoppers --rec "Baudewijn Odé" --alt 1000-2000 --year ">2023" --output_dir ./data

Download frog sounds filtered by sample rate and quality:

xcapi --grp frogs --smp ">44100" --q "<C" --output_dir ./data

Re-run the same command — xcapi will automatically skip recordings already downloaded and only fetch new ones. Note that skipping only works when you point to the same --output_dir as a previous download, and that directory contains either xcapi_runs.json or metadata.csv:

xcapi --grp frogs --smp ">44100" --q "<C" --output_dir ./data

---

Previewing before downloading

Use --metadata_only to check what a query returns and how many recordings you don't have yet, without downloading any audio. Results are saved to metadata_only.csv:

xcapi --grp birds --cnt France --output_dir ./data --metadata_only

This is useful for estimating download size or checking for new additions before committing to a full download.

---

Re-downloading everything from scratch

Use --redownload to ignore previous download records and start completely fresh. This re-downloads all recordings and overwrites metadata.csv and xcapi_runs.json:

xcapi --grp birds --cnt France --output_dir ./data --redownload

---

Python API

from xcapi.query import QueryBuilder
from xcapi.client import XenoCantoClient
from xcapi.downloader import Downloader

# Set up client with your API key
client = XenoCantoClient(api_key="your-api-key")

# Alternatively, use a .env file to avoid passing the key each time:
# Create a file called .env in your working directory and add:
# XENO_CANTO_API_KEY=your-api-key
# Then simply run:
client = XenoCantoClient()

# Build a query
query = QueryBuilder().group("birds").country("Spain").quality("A").build()

# Fetch recording metadata from Xeno-canto
recordings = client.search(query)

# Download recordings
# xcapi will skip recordings already downloaded and only fetch new ones
downloader = Downloader(output_dir="./data")
downloader.download_recordings(recordings)

# Preview what's new without downloading audio
# Writes metadata_only.csv with recordings not yet downloaded
downloader.save_metadata_only(recordings)

# Re-download everything from scratch
downloader.download_recordings(recordings, redownload=True)

---

Available query filters

You can chain multiple filters when building a query:

Taxonomic filters genus(genus_name) species(species_name) subspecies(subspecies_name) family(family_name) group(group_name) – e.g. "birds", "grasshoppers", "bats", "land mammals", "frogs" english_name(name) Geographic filters country(country_name) location(location_name) area(region_name) – e.g. "Europe", "Asia", "Africa" bounding_box(lat_min, lon_min, lat_max, lon_max) latitude(range_or_operator) – e.g. "40-45", ">50" longitude(range_or_operator) – e.g. "-10-0", "<-100" altitude(range_or_operator) – e.g. "100-500", "<1000" Quality and type filters quality(rating) – one of "A", "B", "C", "D", "E", or operators like ">B", "<C" sound_type(type) – e.g. "song", "call", "drumming" sex(sex) – e.g. "male", "female" life_stage(stage) – e.g. "adult", "juvenile" method(method) – e.g. "field recording"

Temporal filters year(year_or_range) – e.g. "2020", "2015-2020", ">2018" month(month_or_range) – e.g. "6", "1-3" since(days) – recordings uploaded in the last N days – e.g. "2", "5" or since a date – e.g. "2012-11-09" time_of_day(time_or_range) – e.g. "06:00", "06:00-12:00" Other and metadata filters recordist(name) – e.g. "Raziya Qadri" length(length_or_range) – e.g. "10-20", "<30", ">60" license(license_type) – e.g. "cc-by", "cc0" also(species) – background species animal_seen(True/False) playback_used(True/False) xc_number(value) – e.g. "76967", "88888-88890", ">76967" temperature(value) – e.g. "20-30", "<10" registration_number(value) automatic_recording("yes"|"no"|"unknown") device(device_name) microphone(microphone_name) sample_rate(rate) – e.g. "44100", ">48000", "44100-96000" remarks(text)

Visit this page for more detail on search tags.