manifest refresh (2026-05-16T13:11:55.442455+00:00)

README.md

@@ -49,7 +49,7 @@ signal bytes live as canonical Zarr stores on S3 and are read on demand.
 | Apples-to-apples compression vs source EEG (rows with `canonical_size_bytes` set) | **3.05×** |
 
 Every canonical row points at a Zarr-v3 group at
-`s3://eeg-datasets-139156132535/canonical/v1/<dataset_id>/<recording_id>.eegz/`
+`s3://eeg-corpus-139156132535/canonical/v1/<dataset_id>/<recording_id>.eegz/`
 with FLAC-compressed `int16` signal + parallel channel / events /
 annotations arrays. Open any one through `manifest.eegz.open_recording`
 or directly with `zarr.open(canonical_uri)` and get the same on-disk
@@ -94,7 +94,7 @@ re-hashing those stores byte-for-byte returns 2,402 / 2,402 matches.
 ```python
 from manifest.eegz import open_recording
 
-uri = "s3://eeg-datasets-139156132535/canonical/v1/hbn_eeg/<recording_id>.eegz/"
+uri = "s3://eeg-corpus-139156132535/canonical/v1/hbn_eeg/<recording_id>.eegz/"
 rec = open_recording(uri)
 print(rec.metadata.n_channels, rec.metadata.sampling_rate_hz, rec.metadata.duration_s)
 
@@ -106,7 +106,7 @@ Or with pure Zarr:
 
 ```python
 import zarr
-store = zarr.open("s3://eeg-datasets-139156132535/canonical/v1/<dataset>/<rid>.eegz/", mode="r")
+store = zarr.open("s3://eeg-corpus-139156132535/canonical/v1/<dataset>/<rid>.eegz/", mode="r")
 signal = store["signal"][:]              # (n_channels, n_samples)
 sfreq  = store.attrs["sampling_rate_hz"]
 chnames = store["channels/name"][:]
@@ -327,8 +327,8 @@ python -m manifest.eegz convert-one \
 
 # Build the published manifest from the per-dataset conversion ledgers + canonical zarr.json.
 python -m manifest.eegz_to_manifest \
-    --input  s3://eeg-datasets-139156132535/manifest/latest/recordings.parquet \
-    --output s3://eeg-datasets-139156132535/manifest/latest/recordings.parquet \
+    --input  s3://eeg-corpus-139156132535/manifest/latest/recordings.parquet \
+    --output s3://eeg-corpus-139156132535/manifest/latest/recordings.parquet \
     --also-publish-latest
 
 # Push the manifest to this Hugging Face dataset repo.