try two methods

src/predict.py

@@ -206,7 +206,7 @@ class Predictor:
         return {"url": file_url}
 
 
-    def predict(self,s3_url,passage,process_audio,method_type='voice_clone'):
+    def predict(self,s3_url,passage,process_audio):
         output_dir = 'processed'
         gen_id = str(uuid.uuid4())
         os.makedirs(output_dir,exist_ok=True)
@@ -223,48 +223,43 @@ class Predictor:
         bucket_name = 'demovidelyuseruploads'
         local_file_path = os.path.join(raw_dir,s3_key)
         self.download_file_from_s3(self.s3_client,bucket_name,s3_key,local_file_path)
-        
-        se_extractor.generate_voice_segments(local_file_path,segments_dir,vad=True)
-
-        if method_type == 'voice_clone':
-            #voice_clone with styletts2
-            model,sampler = self.model,self.sampler
-            processed_seg_dir = os.path.join(segments_dir,s3_key.split('.')[0],'wavs')
-            result = self.process_audio_file(local_file_path,passage,model,sampler)
-            final_output = os.path.join(results_dir,f"{gen_id}-voice-clone-1.wav")
-
-            sf.write(final_output,result,24000)
-            if process_audio:
-                (new_sr, wav1) = self._fn(final_output,"Midpoint",32,0.5)
-                sf.write(final_output,wav1,new_sr)
+        #voice_clone with styletts2
+        model,sampler = self.model,self.sampler
+        result = self.process_audio_file(local_file_path,passage,model,sampler)
+        final_output = os.path.join(results_dir,f"{gen_id}-voice-clone-1.wav")
 
-            base_speaker_tts,tone_color_converter = self.base_speaker_tts,self.tone_color_converter
-            reference_speaker = local_file_path
-            target_se, audio_name = se_extractor.get_se(reference_speaker, tone_color_converter, target_dir=openvoice_dir, vad=False)
-            src_path = os.path.join(results_dir,f"{gen_id}-tmp.wav")
-            openvoice_output = os.path.join(results_dir,f"{gen_id}-voice-clone-2.wav")
-            base_speaker_tts.tts(passage,src_path,speaker='default',language='English',speed=1.0)
-
-            source_se = torch.load(f'{self.ckpt_base}/en_default_se.pth').to(self.device)
-            tone_color_converter.convert(audio_src_path=src_path,src_se=source_se,tgt_se=target_se,output_path=openvoice_output,message='')
-            if process_audio:
-                (new_sr, wav1) = self._fn(openvoice_output,"Midpoint",32,0.5)
-                sf.write(openvoice_output,wav1,new_sr)
-           
-           
-            mp3_final_output_1 = str(final_output).replace('wav','mp3')
-            mp3_final_output_2 = str(openvoice_output).replace('wav','mp3')
-            self.convert_wav_to_mp3(final_output,mp3_final_output_1)
-            self.convert_wav_to_mp3(openvoice_output,mp3_final_output_2)
-            print(mp3_final_output_1)
-            print(mp3_final_output_2)
-            
-            self.upload_file_to_s3(mp3_final_output_1,'demovidelyusergenerations',f"{gen_id}-voice-clone-1.mp3")
-            self.upload_file_to_s3(mp3_final_output_2,'demovidelyusergenerations',f"{gen_id}-voice-clone-2.mp3")
-            shutil.rmtree(os.path.join(output_dir,gen_id))
-            return {"voice_clone_1":f"https://demovidelyusergenerations.s3.amazonaws.com/{gen_id}-voice-clone-1.mp3",
-                    "voice_clone_2":f"https://demovidelyusergenerations.s3.amazonaws.com/{gen_id}-voice-clone-2.mp3"
-                    }
+        sf.write(final_output,result,24000)
+        if process_audio:
+            (new_sr, wav1) = self._fn(final_output,"Midpoint",32,0.5)
+            sf.write(final_output,wav1,new_sr)
+
+        base_speaker_tts,tone_color_converter = self.base_speaker_tts,self.tone_color_converter
+        reference_speaker = local_file_path
+        target_se, audio_name = se_extractor.get_se(reference_speaker, tone_color_converter, target_dir=openvoice_dir, vad=False)
+        src_path = os.path.join(results_dir,f"{gen_id}-tmp.wav")
+        openvoice_output = os.path.join(results_dir,f"{gen_id}-voice-clone-2.wav")
+        base_speaker_tts.tts(passage,src_path,speaker='default',language='English',speed=1.0)
+
+        source_se = torch.load(f'{self.ckpt_base}/en_default_se.pth').to(self.device)
+        tone_color_converter.convert(audio_src_path=src_path,src_se=source_se,tgt_se=target_se,output_path=openvoice_output,message='')
+        if process_audio:
+            (new_sr, wav1) = self._fn(openvoice_output,"Midpoint",32,0.5)
+            sf.write(openvoice_output,wav1,new_sr)
+        
+        
+        mp3_final_output_1 = str(final_output).replace('wav','mp3')
+        mp3_final_output_2 = str(openvoice_output).replace('wav','mp3')
+        self.convert_wav_to_mp3(final_output,mp3_final_output_1)
+        self.convert_wav_to_mp3(openvoice_output,mp3_final_output_2)
+        print(mp3_final_output_1)
+        print(mp3_final_output_2)
+        
+        self.upload_file_to_s3(mp3_final_output_1,'demovidelyusergenerations',f"{gen_id}-voice-clone-1.mp3")
+        self.upload_file_to_s3(mp3_final_output_2,'demovidelyusergenerations',f"{gen_id}-voice-clone-2.mp3")
+        shutil.rmtree(os.path.join(output_dir,gen_id))
+        return {"voice_clone_1":f"https://demovidelyusergenerations.s3.amazonaws.com/{gen_id}-voice-clone-1.mp3",
+                "voice_clone_2":f"https://demovidelyusergenerations.s3.amazonaws.com/{gen_id}-voice-clone-2.mp3"
+                }
         if method_type == 'voice_clone_with_emotions':
             try:
                 print("INSIDE emotions")        
@@ -324,6 +319,50 @@ class Predictor:
             }
     
 
+
+    def predict_with_emotions(self,s3_url,passage,process_audio):
+        output_dir = 'processed'
+        gen_id = str(uuid.uuid4())
+        os.makedirs(output_dir,exist_ok=True)
+        raw_dir = os.path.join(output_dir,gen_id,'raw')
+        segments_dir = os.path.join(output_dir,gen_id,'segments')
+        results_dir = os.path.join(output_dir,gen_id,'results')
+        openvoice_dir = os.path.join(output_dir,gen_id,'openvoice')
+        os.makedirs(raw_dir)
+        os.makedirs(segments_dir)
+        os.makedirs(results_dir)
+        
+        
+        s3_key = s3_url.split('/')[-1]
+        bucket_name = 'demovidelyuseruploads'
+        local_file_path = os.path.join(raw_dir,s3_key)
+        self.download_file_from_s3(self.s3_client,bucket_name,s3_key,local_file_path)       
+        try:
+            print("INSIDE new emotions method")        
+            base_speaker_tts,tone_color_converter = self.base_speaker_tts,self.tone_color_converter
+            reference_speaker = local_file_path
+            print("here 1")
+            target_se, audio_name = se_extractor.get_se(reference_speaker, tone_color_converter, target_dir=openvoice_dir, vad=False)
+            print("here 2")
+            src_path = os.path.join(results_dir,f"{gen_id}-tmp-emotions.wav")
+            openvoice_output = os.path.join(results_dir,f"{gen_id}-4.wav")
+            base_speaker_tts.tts(passage,src_path,speaker='default',language='English',speed=1.0,use_emotions=True)
+            source_se = torch.load(f'{self.ckpt_base}/en_style_se.pth').to(self.device)
+            tone_color_converter.convert(audio_src_path=src_path,src_se=source_se,tgt_se=target_se,output_path=openvoice_output,message='')
+            if process_audio:
+                (new_sr, wav1) = self._fn(openvoice_output,"Midpoint",32,0.5)
+                sf.write(openvoice_output,wav1,new_sr)
+            
+            mp3_final_output_1 = str(openvoice_output).replace('wav','mp3')
+            self.convert_wav_to_mp3(openvoice_output,mp3_final_output_1)
+            print(mp3_final_output_1)
+            self.upload_file_to_s3(mp3_final_output_1,'demovidelyusergenerations',f"{gen_id}-voice-with-emotions.mp3")
+            shutil.rmtree(os.path.join(output_dir,gen_id))
+            return {"voice_clone_with_emotions":f"https://demovidelyusergenerations.s3.amazonaws.com/{gen_id}-voice-with-emotions.mp3"
+                    }
+        except Exception as e:
+            return {"error":f"Unexpected error{e}"}
+
     def _fn(self,path, solver, nfe, tau):
         if path is None:
             return None, None

src/rp_handler.py

@@ -49,7 +49,10 @@ def run_voice_clone_job(job):
         if process_audio is None:
             process_audio = False
 
-        result = MODEL.predict(s3_url,passage,process_audio,method_type)
+        if method_type == 'voice_clone':
+            result = MODEL.predict(s3_url,passage,process_audio)
+        if method_type == 'voice_clone_with_emotions':
+            result = MODEL.predict_with_emotions(s3_url,passage,process_audio)
 
         return result