A place, then a perspective.
The same globe and selected place connect Radio and News. Choose a populated place, adjust the radius and switch lenses. Radio connects directly to broadcasters; news links to original reporting with an original-language summary when the evidence supports one.
Published coverage
34,145 populated places · 13,088 distinct catalog station records · 17 candidate news feeds across 10 pilot places · 13 published local story clusters.
Snapshot checked 2026-09-16T19:00:07.489025+00:00. Place coverage starts with GeoNames cities of at least 15,000 people, supplemented by pilot locations. It does not claim every settlement, language or broadcaster. Radio points come from a community catalog and are not verified transmitter coordinates. Selecting empty terrain resolves the nearest indexed populated-place center.
Tap a place for current headlines
Selecting a place searches GDELT and Google News independently of this catalog. Newly found publisher domains and their matching headline links are saved to the source catalog, along with the place and search evidence. The atlas then checks publisher feeds and shows up to five recent headlines in their original language. Indexed links remain usable if a publisher feed cannot be retrieved. GDELT dates are explicitly labeled as the time first indexed, which can differ from the publication date.
An indexed area match can come from a mention in the article body, and does not verify the publisher’s newsroom location or the article’s local importance. These publishers are labeled as coverage matches and receive no invented distance or map pin. Same-named places can still be confused. Existing sources with known locations retain their distance labels.
Successful searches are reused for six hours; partial searches retry after fifteen minutes and provider failures after five minutes. Rapid concurrent searches may show a brief busy message. Search coverage is incomplete: no returned matches does not establish that an area has no news outlets. Disabled or removed discovered publishers are not automatically restored. This discovery process does not touch the separate source staging queue. Browse the growing catalog · GDELT search documentation.
Tap a headline to inspect its words. English and Japanese have local dictionaries; other languages offer external dictionary and translation links. No automatic grammar analysis or translation is presented as human-verified. Feed results are cached for ten minutes. If current retrieval fails, dated prepared stories can remain available. Publisher restrictions are respected, and missing results are never filled with invented headlines.
Prepared summaries remain evidence-based
The news pilot covers Phoenix, Bucharest, Tokyo, Taipei, Reykjavík, Beirut, Bethel, Whanganui, Cusco and Dakar. A publisher’s location alone never makes an article local. Candidate checks record place mentions, age and source footprint. Editorial review supplies local-consequence and reporting evidence; independently corroborated stories receive an additional score. Government notices stay labeled as government material.
We publish up to five diverse story clusters, with no more than two in one topic. Each summary sentence points to its reporting sources. If access is blocked, evidence is thin or review is incomplete, the place can have fewer than five—or no summaries. Source links remain available. Original headlines are separate from AUWEN summaries. English translations are secondary.
What a radio check means
“Catalog reports working” is Radio Browser’s observation. “AUWEN: endpoint reachable” means a bounded request received recognizable audio bytes or an HLS manifest; it does not establish broadcast identity, listening quality or availability in your country. One failed check does not mean a station closed. The directory-failure filter exposes those reports without deleting station records. HTTP-only streams have external links because the secure embedded player cannot use them. Listener playback never uploads a recording.
Language tools
English lookup uses 147,306 WordNet headwords. Japanese lookup uses 465,565 JMdict spelling and reading keys; these are lookup keys, not that many distinct words. Multiword matching runs before limited inflection lookup. Dictionary senses are candidates, not a promised contextual translation. The reader uses language-aware segmentation, preserves selectable text, supports right-to-left text and allows hover, touch and keyboard inspection.
Princeton WordNet 3.0 · WordNet license · JMdict / EDRDG · CC BY-SA 4.0 terms. JMdict snapshot: 14 September 2026. The derived Japanese dictionary shards retain the same license; AUWEN simplifies their structure and groups lookup keys. Monthly dictionary updates are part of atlas maintenance.
Catalog credits and corrections
Existing radio source explorer · Existing news source explorer. Owner-entered catalogs are preserved alongside the atlas pilot.
GeoNames, used under CC BY 4.0; fields were selected and aliases shortened. Radio Browser supplies station metadata and catalog check history. Stream URLs remain the broadcaster’s; the catalog does not grant rights to rebroadcast or bypass restrictions.
Source metadata, stream endpoints, dated collection observations and summaries are kept separately. Owner corrections can move, merge, suppress and restore records while retaining their original baseline and an audit trail. Owner review desk →
Still developing
Worldwide local-news coverage, community/resident review, full contextual language analysis, automatic semantic clustering across languages, more comprehensive regional dictionaries and independent long-term station-health classification remain ongoing work. No production microphone/audio-quality or iPhone radio test is implied by a successful endpoint check.
Latest source collection observations
KJZZ · en · editorial
2026-09-16T19:00:07.499208+00:00: empty
B365 · ro · editorial
2026-09-16T19:00:07.499812+00:00: ok
Buletin de București · ro · editorial
2026-09-16T19:00:07.501587+00:00: ok
NHK · ja · editorial
2026-09-16T05:15:28.335617+00:00: failed Publisher robots excludes ChatGPT collection; retain directory link only.
Tokyo Shimbun · ja · editorial
2026-09-16T05:15:28.335638+00:00: failed Robots disallows collection
Liberty Times · zh-Hant · editorial
2026-09-16T05:15:28.335960+00:00: failed Robots unavailable: 403
CNA · zh-Hant · editorial
2026-09-16T05:15:28.336294+00:00: failed HTTP Error 404: Not Found
RÚV · is · editorial
2026-09-16T19:00:30.952415+00:00: ok
National News Agency · ar · government
2026-09-16T05:15:47.299939+00:00: failed HTTP Error 404: Not Found
L’Orient-Le Jour · fr · editorial
2026-09-16T05:15:54.097996+00:00: failed Robots disallows collection
KYUK · en · editorial
2026-09-16T19:00:33.720871+00:00: empty
RNZ · en · editorial
2026-09-16T05:16:11.898323+00:00: failed Robots disallows collection
Cusco Post · es · editorial
2026-09-16T05:16:12.719195+00:00: failed Non-public or unresolved address
Diario del Cusco · es · editorial
2026-09-16T19:00:37.789840+00:00: ok
Seneweb · fr · editorial
2026-09-16T05:26:33.859064+00:00: failed HTTP Error 404: Not Found
Dakaractu · fr · editorial
2026-09-16T19:00:55.610586+00:00: ok
シブヤ経済新聞 · ja · editorial
2026-09-16T19:01:00.548972+00:00: ok