32000

app.py

@@ -21,7 +21,7 @@ def map_to_pred(file_path):
     audio, _ = librosa.load(file_path)
 
     # preprocess audio and generate standard
-    input_features = processor([audio], return_tensors="pt", sampling_rate=32_000).input_features
+    input_features = processor([audio], return_tensors="pt", sampling_rate=32000).input_features
     generated_ids = model.generate(inputs=input_features)
     transcription = processor.batch_decode(generated_ids, normalize=True, skip_special_tokens=True)
     text = processor.tokenizer._normalize(transcription[0])