Update README.md

README.md

@@ -147,10 +147,19 @@ GigaSpeech 2
 ```
 
 ### Data Instances
-TODO:
+```shell
+Audio file (.wav):
+    Channels: 1
+    Sample Rate: 16000
+    Sample Encoding: 16-bit Signed Integer PCM
+
+Transcript file (.tsv):
+    <segment_id>\t<text>\n
+```
 
 ### Data Fields
-TODO:
+- segment_id (string) - string id of the segment.
+- text (string) - transcription of the segment.
 
 ### Data Splits
 The dataset has three subsets: train, dev and test.