Update app.py

app.py

@@ -1,13 +1,19 @@
 import streamlit as st
 from transformers import pipeline
 from transformers import T5Config
+from datasets import load_dataset
+
+embeddings_dataset = load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")
+
+speaker_embeddings = embeddings_dataset[7306]["xvector"]
+speaker_embeddings = torch.tensor(speaker_embeddings).unsqueeze(0)
+
 
 # 加载 Visual Question Answering 模型 microsoft/git-base-vqav2
 vqa_pipeline = pipeline("text2text-generation", model="microsoft/git-base-vqav2")
 
-# 加载文本到语音模型并手动指定 speaker_embeddings
-config = T5Config.from_pretrained("microsoft/speecht5_tts", speaker_embeddings=True)
-text_to_speech_pipeline = pipeline("text-to-speech", model="microsoft/speecht5_tts", config=config)
+# 加载文本到语音模型并手动指定
+text_to_speech_pipeline = pipeline("text-to-speech", model="microsoft/speecht5_tts")
 
 def main():
     st.title("Visual Question Answering with Text-to-Speech")