Update generate_audio.py

generate_audio.py

@@ -38,15 +38,15 @@ class TTSGenerator:
         self.speaker1_description = """
         Laura's voice is expressive and dramatic in delivery, speaking at a moderately fast pace with a very close recording that almost has no background noise.
         """
-         self.speaker2_description = """
+        self.speaker2_description = """
         Gary's voice is expressive and dramatic in delivery, speaking at a moderately fast pace with a very close recording that almost has no background noise.
         """
         
         # Load Bark model and processor for Speaker 2
-        self.bark_processor = AutoProcessor.from_pretrained("suno/bark")
-        self.bark_model = BarkModel.from_pretrained("suno/bark", torch_dtype=torch.float16).to(self.device)
-        self.bark_sampling_rate = 24000
-        self.voice_preset = "v2/en_speaker_6"
+        # self.bark_processor = AutoProcessor.from_pretrained("suno/bark")
+        # self.bark_model = BarkModel.from_pretrained("suno/bark", torch_dtype=torch.float16).to(self.device)
+        # self.bark_sampling_rate = 24000
+        # self.voice_preset = "v2/en_speaker_6"
 
     @spaces.GPU
     def load_transcript(self):
@@ -89,10 +89,16 @@ class TTSGenerator:
             np.array: Audio array.
             int: Sampling rate.
         """
-        inputs = self.bark_processor(text, voice_preset=self.voice_preset).to(self.device)
-        speech_output = self.bark_model.generate(**inputs, temperature=0.9, semantic_temperature=0.8)
-        audio_arr = speech_output[0].cpu().numpy()
-        return audio_arr, self.bark_sampling_rate
+
+         input_ids = self.parler_tokenizer(self.speaker2_description, return_tensors="pt").input_ids.to(self.device)
+        prompt_input_ids = self.parler_tokenizer(text, return_tensors="pt").input_ids.to(self.device)
+        generation = self.parler_model.generate(input_ids=input_ids, prompt_input_ids=prompt_input_ids)
+        audio_arr = generation.cpu().numpy().squeeze()
+        return audio_arr, self.parler_model.config.sampling_rate
+        # inputs = self.bark_processor(text, voice_preset=self.voice_preset).to(self.device)
+        # speech_output = self.bark_model.generate(**inputs, temperature=0.9, semantic_temperature=0.8)
+        # audio_arr = speech_output[0].cpu().numpy()
+        # return audio_arr, self.bark_sampling_rate
 
     @staticmethod
     @spaces.GPU