Update app.py

app.py

@@ -20,9 +20,9 @@ huggingface_token = os.environ["huggingface_token"]
 pipe = pipeline(model="mskov/whisper-small-esc50") 
 print(pipe)
 processor = WhisperProcessor.from_pretrained("mskov/whisper-small-esc50")
-dataset = load_dataset("mskov/misophonia_sounds", split="test").cast_column("audio", Audio(sampling_rate=16000))
+dataset = load_dataset("ashraq/esc50", split="train").cast_column("audio", Audio(sampling_rate=16000))
 
-print(dataset, "and at 0[audio][array] ", dataset[0]["audio"]["array"], type(dataset[0]["audio"]["array"]), "and at audio : ", dataset[0]["audio"])
+# print(dataset, "and at 0[audio][array] ", dataset[0]["audio"]["array"], type(dataset[0]["audio"]["array"]), "and at audio : ", dataset[0]["audio"])
 
 model = WhisperForConditionalGeneration.from_pretrained("mskov/whisper-small-esc50")
 
@@ -35,14 +35,14 @@ model = WhisperForConditionalGeneration.from_pretrained("mskov/whisper-small-esc
 
     
 def map_to_pred(batch):
-    cleaned_transcription = re.sub(r'\[[^\]]+\]', '', batch['sentence']).strip()
+    cleaned_transcription = re.sub(r'\[[^\]]+\]', '', batch['category']).strip()
     print("cleaned transcript", cleaned_transcription)
-    cleaned_transcription = preprocess_transcription(batch['sentence'])
+    cleaned_transcription = preprocess_transcription(batch['category'])
     normalized_transcription = processor.tokenizer._normalize(cleaned_transcription)
 
     audio = batch["audio"]
     input_features = processor(audio["array"], sampling_rate=audio["sampling_rate"], return_tensors="pt").input_features
-    batch["reference"] = processor.tokenizer._normalize(batch['sentence'])
+    batch["reference"] = processor.tokenizer._normalize(batch['category'])
     
     
     with torch.no_grad():