change cosyvoice/cli/frontend.py use cpu provider for onnx to avoid zero gpu error

cosyvoice/cli/frontend.py

@@ -42,7 +42,9 @@ class CosyVoiceFrontEnd:
         option.graph_optimization_level = onnxruntime.GraphOptimizationLevel.ORT_ENABLE_ALL
         option.intra_op_num_threads = 1
         self.campplus_session = onnxruntime.InferenceSession(campplus_model, sess_options=option, providers=["CPUExecutionProvider"])
-        self.speech_tokenizer_session = onnxruntime.InferenceSession(speech_tokenizer_model, sess_options=option, providers=["CUDAExecutionProvider"])
+        # self.speech_tokenizer_session = onnxruntime.InferenceSession(speech_tokenizer_model, sess_options=option, providers=["CUDAExecutionProvider"])
+        # use cpu provider for onnx to avoid zero gpu error
+        self.speech_tokenizer_session = onnxruntime.InferenceSession(speech_tokenizer_model, sess_options=option, providers=["CPUExecutionProvider"])
         if os.path.exists(spk2info):
             self.spk2info = torch.load(spk2info, map_location=self.device)
         self.instruct = instruct