add word detection times to dataset

audio-kw-in-context.py

@@ -61,6 +61,9 @@ class AudioKwInContext(datasets.GeneratorBasedBuilder):
                     "sentence": datasets.Value("string"),
                     "keywords": datasets.Sequence(datasets.Value("string")),  
                     "audio": datasets.Audio(sampling_rate=16_000),
+                    "start_times": datasets.Sequence(datasets.Value("float32")),
+                    "end_times": datasets.Sequence(datasets.Value("float32")),
+                    "confidence": datasets.Sequence(datasets.Value("float32")),
                 }
         )
 
@@ -132,6 +135,9 @@ class AudioKwInContext(datasets.GeneratorBasedBuilder):
                     "speaker_id": row['client_id'],
                     "keywords": row['keywords'],  
                     "audio": tfile,
+                    "start_times": row['start_times'],
+                    "end_times": row['end_times'],
+                    "confidence": row['confidence'],
                 }
             except Exception as e:
                 print(e)