Document canonical FormosanBank audio provenance

README.md

@@ -2,7 +2,11 @@
 license: other
 license_name: formosanbank-terms-ai-use-addendum
 license_link: https://ai4commsci.gitbook.io/formosanbank/additional-resources/terms-of-use
+task_categories:
+- automatic-speech-recognition
 tags:
+- audio
+- formosan-languages
 - noncommercial
 - no-commercial-ai
 ---
@@ -11,7 +15,19 @@ tags:
 
 > Commercial AI Use is prohibited without prior written permission. See the FormosanBank Terms of Use and AI Use Addendum.
 
-Dataset card for FormosanBank/ePark_yue_du_shu_xie_pian_reading_writing.
+This is a canonical public audio dataset for the FormosanBank corpus.
+
+## Provenance and inventory
+
+- Public corpus source: [Corpora/ePark/XML/yue_du_shu_xie_pian_reading_writing](https://github.com/FormosanBank/FormosanBank/tree/98df3773bf26ae2ef5373cd97e0841762126e61d/Corpora/ePark/XML/yue_du_shu_xie_pian_reading_writing)
+- Source baseline commit: [`98df3773bf26ae2ef5373cd97e0841762126e61d`](https://github.com/FormosanBank/FormosanBank/commit/98df3773bf26ae2ef5373cd97e0841762126e61d)
+- Canonical audio files: 30,780
+- Duration: 475.68 hours across the complete 11-repository ePark corpus (corpus statistics)
+- Audio form: Public source recordings.
+- Clone destination: `Corpora/ePark/Audio/yue_du_shu_xie_pian_reading_writing`
+- Declared public extras: 157 files retained by `audio_extras.json`
+
+The pinned revision and exact XML-to-audio contract are recorded in [`audio_sources.json`](https://github.com/FormosanBank/FormosanBank/blob/main/audio_sources.json). Download this corpus with its `download_audio_data.sh` wrapper or download all public audio with `run_audio_downloads.sh`.
 
 ## License and AI Use