Radiogarden API used to create stations.json list with +27,900 stations

[johnmduffy76]

I've been enjoying my RadioGlobe for the past 8 months with about 10,000 loaded stations but I was always bothered by annoying dead-zones here and there. Therefore, I wanted to try to incorporate all of the stations listed on the RadioGarden website since this is a vast collection of streaming sites. Over a few weeks I worked on this goal and I've gotten the station.json file to the point where I'm willing to share it. The process was as follows:

1. Radiogarden lists stations in cities with a "places" code of which there are about 12,500 of them. The "places" codes were able to be collected here: http://radio.garden/api/ara/content/places
2. Using a script, I was able to examine each "places" code using a prefix, and collect all radio stations within that place. This step resulted with over 150,000 stations...but only returned the "station" code...not the actual streaming website URL. Getting the actual URLs took another step. The 150,000 number is a bit misleading...many stations are used in multiple location (eg. BBC World Service). There are at least 27,940 distinct stations in this new list.
3. Examine all 12,500 cities and determine their lat/lon coordinates...save to a table.
4. Using another script I was able to look at each individual station code website and see where it 304 re-directed...then save this URL.
5. Use another script to create a database which combines the data from steps 4 & 5.
6. Another script to identify the 2-digit standardized country code and apply to the end of each city name appropriately. Some mistakes happened here which I'm still working to fix. Many cities around the world use the same city name and the script made some errors. I'm trying to create another script to look at the lat/lon coords and correlate that the country code is correct...still INW.
7. Clean up the data, remove non-US characters, shorten station titles to 16-characters, etc, etc.

The station list works fine, although it does create a few new problems. Because now there are so many cities, it becomes difficult to tune to a specific city and instead the globe will "snap" to a nearby city. Adjusting the snap distance would work, but make it overly sensitive. I would like to implement a new "city select" function which might help with this. The user should be able to use the jog wheel to move through nearby cities. More to come on this function.

Also, I have had a few lock-ups of the RadioGlobe here and there where the screen will not respond and the globe won't move to a new station (although it will continue playing the current station). This may be related to the code having an issue with too many stations in one place but this is only a theory.

-John
stations.json

[johnmduffy76]

Finished 150k stations (v1).xlsx

[vfinel]

thank you so much @johnmduffy76 !
I've created a pull request so the updated file could be added to the repository

Would you be willing to share the code you've used to fetch the new stations ? The code sounds cool and it would be valuable for future updates, shall the current list get outdated in a few years

[johnmduffy76]

Hello vfinel,

I'll provide the code here and lay out my process to get the RadioGarden API data onto the RadioGlobe. This could be repeated yearly and will keep the Globe up-to-date.

STEP 1:
(radio_garden_processor.py) below will look at Radiogarden's API and build an Excel table with 5 columns of data. It will indicate all failed links which could be re-examined again later at another time of day. It also reformats all non-Latin characters and translates the titles to English. The Excel table format can then be easily modified by users at this step if they desire.

---

import requests
import pandas as pd
import json
import time
import pycountry
import re
from unidecode import unidecode # For transliteration

File paths

input_file = '/Users/john/Desktop/radio_garden_places.txt'
output_file = '/Users/john/Desktop/RadioGarden_Complete_Data.xlsx'

Base API URLs

location_api_url = "https://radio.garden/api/ara/content/page/"
channel_api_url = "https://radio.garden/api/ara/content/channel/"
stream_api_url = "https://radio.garden/api/ara/content/listen/{channel_id}/channel.mp3"

Load and fix the places data

with open(input_file, 'r', encoding='utf-8') as f:
json_str = f.read().strip()
if not json_str.endswith('}}}'):
last_valid = json_str.rfind('},')
if last_valid != -1:
json_str = json_str[:last_valid + 1] + ']}}'
else:
raise ValueError("Cannot find a valid JSON structure to fix")
places_data = json.loads(json_str)

Initialize lists

names = []
latitudes = []
longitudes = []
station_names = []
stream_urls = []

Counter for processed stations

station_count = 0

Function to transliterate and clean text

def clean_text(text):
# Transliterate non-Latin to Latin (e.g., "Русское" -> "Russkoe")
transliterated = unidecode(text)
# Keep only alphanumeric, spaces, commas, and hyphens
cleaned = ''.join(c for c in transliterated if c.isalnum() or c in ' ,-')
# Remove extra spaces and ensure not empty
cleaned = cleaned.strip()
return cleaned if cleaned else "Unknown Station"

Function to get 2-letter country code

def get_country_code(country_name):
try:
country = pycountry.countries.search_fuzzy(country_name)[0]
return country.alpha_2
except LookupError:
return "XX"

Function to format the "Name" field

def format_name(place_title, country):
country_code = get_country_code(country)
if country_code == "US":
match = re.search(r'\b([A-Z]{2})\b$', place_title)
if match:
state_code = match.group(1)
city = place_title.replace(f" {state_code}", "").strip()
formatted = f"{city},US-{state_code}"
else:
formatted = f"{place_title},US"
else:
formatted = f"{place_title},{country_code}"
return clean_text(formatted)

Function to save progress to Excel (overwrites each time)

def save_progress():
combined_df = pd.DataFrame({
'Name': names,
'Value.coords.n': latitudes,
'Value.coords.e': longitudes,
'Value.urls.name': station_names,
'Value.urls.url': stream_urls
})
combined_df.to_excel(output_file, index=False)
print(f"Saved {len(station_names)} stations to {output_file} (progress update)")

try:
for place in places_data['data']['list']:
location_code = place['id']
place_title = place.get('title', 'Unknown Place')
country = place.get('country', 'Unknown Country')
geo = place.get('geo', [None, None])
longitude, latitude = geo[0], geo[1]
name = format_name(place_title, country)
location_url = f"{location_api_url}{location_code}"

    print(f"Processing: {name} ({location_code})")

    try:
        response = requests.get(location_url, timeout=10)
        response.raise_for_status()
        location_data = response.json()

        content = location_data.get('data', {}).get('content', [])
        stations_found = False
        for content_block in content:
            if content_block.get('itemsType') == 'channel':
                stations_found = True
                for item in content_block.get('items', []):
                    station_title = item.get('title', 'Unknown Station')
                    station_code = item.get('page', {}).get('url', '').split('/')[-1]

                    # Fetch station details
                    channel_url = f"{channel_api_url}{station_code}"
                    try:
                        channel_response = requests.get(channel_url, timeout=10)
                        channel_response.raise_for_status()
                        channel_data = channel_response.json()
                        station_name = channel_data.get('data', {}).get('title', station_title)
                        station_name = clean_text(station_name)  # Clean station name
                    except requests.exceptions.RequestException as e:
                        print(f"  Failed details for {station_code}: {e}")
                        station_name = 'Failed'

                    # Fetch stream URL
                    stream_url = stream_api_url.format(channel_id=station_code)
                    try:
                        stream_response = requests.head(stream_url, allow_redirects=False, timeout=10)
                        live_stream_url = stream_response.headers.get('Location', 'No Stream URL Found') if stream_response.status_code == 302 else 'No Stream URL Found'
                    except requests.exceptions.RequestException as e:
                        print(f"  Failed stream for {station_code}: {e}")
                        live_stream_url = 'Failed'

                    # Store data
                    names.append(name)
                    latitudes.append(latitude)
                    longitudes.append(longitude)
                    station_names.append(station_name)
                    stream_urls.append(live_stream_url)

                    station_count += 1
                    print(f"  Added: {station_name} (Station #{station_count})")

                    # Save every 50 stations
                    if station_count % 50 == 0:
                        save_progress()

                    time.sleep(0.2)

        if not stations_found:
            print(f"  No stations found for {name}")

    except requests.exceptions.RequestException as e:
        print(f"Failed fetching stations for {location_code}: {e}")

    time.sleep(0.2)

except KeyboardInterrupt:
print("\nInterrupted. Saving progress...")
save_progress()

finally:
save_progress()

[johnmduffy76]

STEP 2:

1. Take the output Excel File, and export it to CSV format in Excel.
2. Re-name this file to rg_stations.csv
3. Run the code below (csv_to_json.py) which will format the CSV format to JSON which the RadioGlobe can understand.
4. Upload to the RadioGlobe. Done!

*One issue here is that this creates a .json file with many, many more city locations than the original .json file that the development team originally created. Consolidating these together within a nearby large town/city is probably preferred to having a plethora of small towns where you only see a few stations. If this is the best way forward, it still needs to be done.

---

import pandas as pd
import json
import os

Read the CSV file

df = pd.read_csv('/Users/john/Desktop/RG_stations.csv') # Replace with the path to your CSV file

Group by 'Name' and aggregate into the desired JSON structure

grouped = df.groupby('Name').apply(
lambda x: {
'coords': {
'n': x['Value.coords.n'].iloc[0],
'e': x['Value.coords.e'].iloc[0]
},
'urls': [
{
'name': row['Value.urls.name'],
'url': row['Value.urls.url']
}
for _, row in x.iterrows()
]
}
).to_dict()

Convert the dictionary to JSON

json_data = {k: v for k, v in grouped.items()}

Get the path to the desktop

desktop_path = os.path.join(os.path.expanduser("~"), "Desktop")

Write the JSON data to a file on the desktop

output_file_path = os.path.join(desktop_path, 'stations.json')
with open(output_file_path, 'w') as f:
json.dump(json_data, f, indent=2)

print("JSON file has been created successfully on your desktop")

[vfinel]

Hi @johnmduffy76

Awesome, thank you so much ! I have created a github repository to host your code (making sure to give credits to you ;))
https://github.com/vfinel/RadioGLobeStations

I've also added instructions and some modifications to be able to run the code directly on the globe's raspberry pi.

I will update the stations.json file of my fork of the radio globe software once in a while so that people can directly use it, and submit a merge request to the original repo so that the stations.json keeps up-to-date :) (at least, as up-to-date as mine...)

Nice collaborative work !

[Frank-Bemelman]

Hello @johnmduffy76 - many thanks for this elaborate stations.json file.

I am building my own version of the Radio Globe, using a relative small ESP32 controller. I am using your file to pre-sort the stations by coordinates and store the urls in smaller files. The ESP32 has not enough memory to load it all in one go.

By sheer coincidence I found that...

"Aalst,NL": {
"coords": {
"n": 50.9378,
"e": 4.041
},

and
"Aalst,BE": {
"coords": {
"n": 50.9378,
"e": 4.041
},

...have the same coordinates.

Aalst, NL should be more like "n": 51.783535, ë": 5.129255

Anyway, this is a great resource - thanks again for all that good work!

Cheers,
Frank

[johnmduffy76]

Hi Frank, The most recent version of the "stations.json" file I've created is stored here: https://drive.google.com/drive/u/0/folders/1_tm-YndN73BT1MXVYY2gfjW82Xi9321W It contains 20,597 distinct station locations, but it does not contain the city Aalst, BE or Aalst, NL. Which file are you working with? Something from Github? Good luck with your build. Is there any specific reason you're using an ESP32? Cost? Are you changing the physical design at all? I'd love to see it when you're done! The RadioGlobe is a fantastic project and I still use mine almost every day (it's playing Afrika-Boom - Windhoek, Namibia right now!) Cheers, John

…

On Mon, Jun 9, 2025 at 3:27 PM Frank Bemelman ***@***.***> wrote: *Frank-Bemelman* left a comment (DesignSparkRS/RadioGlobe#7) <#7 (comment)> Hello @johnmduffy76 <https://github.com/johnmduffy76> - many thanks for this elaborate stations.json file. I am building my own version of the Radio Globe, using a relative small ESP32 controller. I am using your file to pre-sort the stations by coordinates and store the urls in smaller files. The ESP32 has not enough memory to load it all in one go. By sheer coincidence I found that... "Aalst,NL": { "coords": { "n": 50.9378, "e": 4.041 }, and "Aalst,BE": { "coords": { "n": 50.9378, "e": 4.041 }, ...have the same coordinates. Aalst, NL should be more like "n": 51.783535, ë": 5.129255 Anyway, this is a great resource - thanks again for all that good work! Cheers, Frank — Reply to this email directly, view it on GitHub <#7 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BG45XN2AQGBJ222ELEZ53M33CWDVNAVCNFSM6AAAAABSXJTVNOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDSNJVG44TMMBXHA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[Frank-Bemelman]

Well, I used the stations.json that you attached to your first message here, with the 150k stations.. Yes, a total redesign. I use the ESP32 as I am much more familiar with that. Indeed, such a wonderful project, lots to learn too. Maybe I should use your latest version, as 150K is quite a collection to handle. I am preparing a repository, but it is far from finished....