Update README.md

README.md

@@ -7,19 +7,25 @@ configs:
   version: 1.0.0
   data_files: data/models.jsonl
 - config_name: datasets
+  version: 1.0.0
+  data_files: data/datasets.jsonl
+- config_name: dataset_utterances
   version: 1.0.0
   data_files:
   - split: aru
-    path: data/aru/files*.jsonl
+    path: data/aru/utterances*.jsonl
   - split: cmu_arctic
-    path: data/cmu-arctic/files*.jsonl
-- config_name: embeddings
+    path: data/cmu-arctic/utterances*.jsonl
+- config_name: utterance_embeddings
   version: 1.0.0
   data_files:
   - split: aru
-    path: data/aru/embeddings*.jsonl
+    path: data/aru/utterance_embeddings*.jsonl
   - split: cmu_arctic
-    path: data/cmu-arctic/embeddings*.jsonl
+    path: data/cmu-arctic/utterance_embeddings*.jsonl
+- config_name: speaker_embeddings
+  version: 1.0.0
+  data_files: data/*/speaker_embeddings*.jsonl
 dataset_info:
 - config_name: speakers
   features:
@@ -49,11 +55,17 @@ dataset_info:
     dtype: string
   - name: name
     dtype: string
-  - name: speaker
+  - name: sample_rate
+    dtype: int32
+- config_name: dataset_utterances
+  features:
+  - name: id
+    dtype: string
+  - name: name
     dtype: string
-  - name: dataset
+  - name: speaker
     dtype: string
-- config_name: embeddings
+- config_name: utterance_embeddings
   features:
   - name: speaker_id
     dtype: string
@@ -63,6 +75,16 @@ dataset_info:
     dtype: string
   - name: embedding
     sequence: float32
+- config_name: speaker_embeddings
+  features:
+  - name: speaker_id
+    dtype: string
+  - name: dataset_id
+    dtype: string
+  - name: model_id
+    dtype: string
+  - name: embedding
+    sequence: float32
 
 pretty_name: Speaker Embeddings
 ---

data/aru/speaker_embeddings.jsonl

@@ -0,0 +1,2 @@
+{ "speaker_id": "001", "dataset_id": "001", "model_id": "001", "embedding": [1.0, 2.0, 3.0] }
+{ "speaker_id": "001", "dataset_id": "001", "model_id": "001", "embedding": [1.0, 2.0, 3.0] }

data/cmu-arctic/speaker_embeddings.jsonl

@@ -0,0 +1,2 @@
+{ "speaker_id": "002", "dataset_id": "002", "model_id": "001", "embedding": [1.0, 2.0, 3.0] }
+{ "speaker_id": "002", "dataset_id": "002", "model_id": "001", "embedding": [1.0, 2.0, 3.0] }

data/datasets.jsonl

@@ -0,0 +1,2 @@
+{ "id": "001", "name": "aru", "sample_rate": 16000 }
+{ "id": "002", "name": "cmu-arctic", "sample_rate": 16000 }