make demo faster

app.py

@@ -33,7 +33,7 @@ def inference(wav,data):
             # Decoding parameters are not included in the model file
             lang_prompt_token="<|en|> <|ner|> <|SLURP|>",
             prompt_token_file="UniverSLU-17-Task-Specifier/add_tokens-Copy1.txt",
-            beam_size=20,
+            beam_size=1,
             ctc_weight=0.0,
             penalty=0.1,
             nbest=1
@@ -60,6 +60,7 @@ def inference(wav,data):
             lang_prompt_token="<|en|> <|ic|> <|fsc|>",
             prompt_token_file="UniverSLU-17-Task-Specifier/add_tokens-Copy1.txt",
             ctc_weight=0.0,
+            beam_size=1,
             nbest=1
           )
           nbests = speech2text(speech)
@@ -78,6 +79,7 @@ def inference(wav,data):
             lang_prompt_token="<|en|> <|ic|> <|SNIPS|>",
             prompt_token_file="UniverSLU-17-Task-Specifier/add_tokens-Copy1.txt",
             ctc_weight=0.0,
+            beam_size=1,
             nbest=1
           )
           nbests = speech2text(speech)
@@ -93,7 +95,7 @@ def inference(wav,data):
             lang_prompt_token="<|nl|> <|scr|> <|grabo_scr|>",
             prompt_token_file="UniverSLU-17-Task-Specifier/add_tokens-Copy1.txt",
             ctc_weight=0.0,
-            beam_size=20,
+            beam_size=1,
             nbest=1
           )
           nbests = speech2text(speech)
@@ -289,7 +291,7 @@ def inference(wav,data):
             lang_prompt_token="<|en|> <|sp|> <|STOP|>",
             prompt_token_file="UniverSLU-17-Task-Specifier/add_tokens-Copy1.txt",
             ctc_weight=0.0,
-            beam_size=20,
+            beam_size=1,
             penalty=0.1,
             nbest=1
           )