Update app.py

app.py

@@ -6,32 +6,24 @@ from fairseq.models.speech_to_speech.hub_interface import S2SHubInterface
 from fairseq.models.speech_to_text.hub_interface import S2THubInterface
 from audio_pipe import SpeechToSpeechPipeline
 
-io1 = gr.Interface.load("huggingface/facebook/xm_transformer_s2ut_en-hk", api_key=os.environ['api_key'])
-io2 = gr.Interface.load("huggingface/facebook/xm_transformer_s2ut_hk-en", api_key=os.environ['api_key'])
-io3 = gr.Interface.load("huggingface/facebook/xm_transformer_unity_en-hk", api_key=os.environ['api_key'])
-io4 = gr.Interface.load("huggingface/facebook/xm_transformer_unity_hk-en", api_key=os.environ['api_key'])
-pipe = SpeechToSpeechPipeline("facebook/xm_transformer_unity_hk-en")
-
-def call_model(audio, model):
-    # pipe = SpeechToSpeechPipeline("facebook/xm_transformer_unity_hk-en")
-    # wav, sr, text = pipe(audio)
-    # temp_file = pipe(audio)
-    # return gr.Audio(temp_file)
-    print(pipe(audio).get_config())
-    return pipe(audio).get_config()["value"]["name"]
+# io1 = gr.Interface.load("huggingface/facebook/xm_transformer_s2ut_en-hk", api_key=os.environ['api_key'])
+# io2 = gr.Interface.load("huggingface/facebook/xm_transformer_s2ut_hk-en", api_key=os.environ['api_key'])
+# io3 = gr.Interface.load("huggingface/facebook/xm_transformer_unity_en-hk", api_key=os.environ['api_key'])
+# io4 = gr.Interface.load("huggingface/facebook/xm_transformer_unity_hk-en", api_key=os.environ['api_key'])
+pipe1 = SpeechToSpeechPipeline("facebook/xm_transformer_s2ut_en-hk")
+pipe2 = SpeechToSpeechPipeline("facebook/xm_transformer_s2ut_hk-en")
+pipe3 = SpeechToSpeechPipeline("facebook/xm_transformer_unity_en-hk")
+pipe4 = SpeechToSpeechPipeline("facebook/xm_transformer_unity_hk-en")
 
 def inference(audio, model):
     if model == "xm_transformer_s2ut_en-hk":
-        out_audio = io1(audio)
+        out_audio = pipe1(audio).get_config()["value"]["name"]
     elif model == "xm_transformer_s2ut_hk-en":
-        out_audio = io2(audio)
+        out_audio = pipe2(audio).get_config()["value"]["name"]
     elif model == "xm_transformer_unity_en-hk":
-        out_audio = io3(audio)
-    elif model == "xm_transformer_unity_hk-en_gpu":
-        out_audio = call_model(audio, model)
+        out_audio = pipe3(audio).get_config()["value"]["name"]
     else:
-        out_audio = io4(audio)
-        print(out_audio)
+        out_audio = pipe4(audio).get_config()["value"]["name"]
     return out_audio