Convert metadata to parquet; update README

README.md

@@ -24,6 +24,30 @@ This dataset is a modified version of the [LibriSpeech](http://www.openslr.org/1
 - **Audio Sampling Rate**: 16 kHz, consistent with LibriSpeech specifications.
 - **Speaker Representation**: Multiple speakers retained from LibriSpeech’s corpus, offering diverse voice samples for training and evaluation.
 
+```
+mumospee_librispeech/
+├── librispeech-parquet/     # audio + transcripts, one parquet per LibriSpeech subset
+│   ├── dev-clean.parquet ... train-other-500.parquet   (7 subsets)
+└── metadata/                # lightweight metadata index, one parquet per subset
+    ├── dev-clean.parquet ... train-other-500.parquet
+```
+
+The `metadata/` parquet files index the audio segments and contain the following columns:
+
+| column | type | description |
+| --- | --- | --- |
+| `path` | string | audio file name (e.g. `2277-149896-0000.flac`) |
+| `url` | string | download URL of the audio parquet shard containing this segment |
+| `type` | string | always `audio` |
+| `duration` | float64 | segment duration in seconds |
+| `language` | string | always `en` |
+| `transcript` | string | transcription |
+| `tag` | string | always `Librispeech` |
+| `split` | string | `train` / `test` / `validation` (LibriSpeech `dev-*` subsets map to `validation`) |
+| `license` | string | always `CC-BY-4.0` |
+
+Total segments: 292,367.
+
 ### Usage
 
 This dataset is well-suited for ASR tasks, optimized for high-throughput training in distributed environments.
@@ -31,3 +55,10 @@ This dataset is well-suited for ASR tasks, optimized for high-throughput trainin
 ### Attribution
 
 This dataset is derived from the original [LibriSpeech](http://www.openslr.org/12) dataset, with modifications to format. Please cite the original LibriSpeech dataset in any publications or projects using this dataset.
+
+### Changelog
+
+#### 2026-06-30
+
+- Converted the `metadata/` index from CSV to parquet (one file per subset).
+- Replaced a leaked local cache path in the audio parquet `file` column with the audio file's basename.

metadata/dev-clean.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:037759305572ae511be81480d9d924c50af2e5c121e9c27bfc69472b2c2f9282
+size 153918

metadata/dev-other.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4e788b9d7721aaae36c2bce1b6b5e26d1d692d1834b5fd3c76add046b6637eb
+size 144378

metadata/test-clean.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d42847537552bfed914e93e7382bdcfc8fd162a5d1823da1c2c53bc18db62fec
+size 149171

metadata/test-other.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:93e027e5209ab843525429531ad78aa462bbe323fac875f9180aa11f1a480914
+size 147221

metadata/train-clean-100.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9dfce5edbaebf1a73be94d46cd1bd54d85ae95f14d8b9e9df95271ae00aa1231
+size 2351133

metadata/train-clean-360.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e28022b66140c46c954840eb266d2a6f4794ac2290ae75a3001cf2e3c2426c4f
+size 8358775

metadata/train-other-500.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fed5e27a884bd218834e02d4bd646a63f6fe998842203b1429676f6b18abe42a
+size 11147838