Update app.py

app.py

@@ -23,10 +23,47 @@ import ffmpeg
 
 block = gr.Blocks()
 
+def one_shot(image,input_text,gender): 
+   return input_text,gender
+   exit()
+    
+   if gender == 'Female' or gender == 'female':
+     tts = gTTS(input_text)
+     with tempfile.NamedTemporaryFile(suffix='.mp3', delete=False) as f:
+          tts.write_to_fp(f)
+          f.seek(0)
+          sound = AudioSegment.from_file(f.name, format="mp3")
+          sound.export("/content/audio.wav", format="wav")
+     one_shot_talking(image,'audio.wav') 
+       
+   elif gender == 'Male' or gender == 'male':
+      print(gender)
+      models, cfg, task = load_model_ensemble_and_task_from_hf_hub(
+          "Voicemod/fastspeech2-en-male1",
+          arg_overrides={"vocoder": "hifigan", "fp16": False}
+      )
+
+      model = models[0].cuda()
+      TTSHubInterface.update_cfg_with_data_cfg(cfg, task.data_cfg)
+      generator = task.build_generator([model], cfg)
+      # next(model.parameters()).device
+
+      sample = TTSHubInterface.get_model_input(task, input_text)
+      sample["net_input"]["src_tokens"] = sample["net_input"]["src_tokens"].cuda()
+      sample["net_input"]["src_lengths"] = sample["net_input"]["src_lengths"].cuda()
+      sample["speaker"] = sample["speaker"].cuda()
+
+      wav, rate = TTSHubInterface.get_prediction(task, model, generator, sample)
+      # soundfile.write("/content/audio_before.wav", wav, rate)
+      soundfile.write("/content/audio_before.wav", wav.cpu().clone().numpy(), rate)
+      cmd='ffmpeg -i /content/audio_before.wav -filter:a "atempo=0.7" -vn /content/audio.wav'
+      os.system(cmd)
+      one_shot_talking(image,'audio.wav')
+
+       
 
 
 def generate_ocr(method,image,gender):
-    print("efef")
     return "Hello"
     
 def run():
@@ -44,7 +81,7 @@ def run():
         with gr.Row().style(equal_height=True):
           btn = gr.Button("Generate")          
 
-    btn.click(generate_ocr, inputs=[image_in, input_text,gender], outputs=[video_out])
+    btn.click(one_shot, inputs=[image_in, input_text,gender], outputs=[video_out])
     # block.queue()
     block.launch(server_name="0.0.0.0", server_port=7860)