Delete web_app

web_app/.env

@@ -1,4 +0,0 @@
-AZURE_OPENAI_ENDPOINT=https://test111222333.openai.azure.com/openai/deployments/gpt-4o/chat/completions?api-version=2025-01-01-preview
-AZURE_OPENAI_API_KEY=8iMbEYWnZI0tYrCLDk1GNUJXPp3VMCMUvl8tdbiVxi1v34vhnI7sJQQJ99AKACfhMk5XJ3w3AAABACOGNIPC
-AZURE_OPENAI_API_VERSION=2024-05-01-preview
-AZURE_OPENAI_DEPLOYMENT=gpt-4o

web_app/.streamlit/config.toml

@@ -1,5 +0,0 @@
-[server]
-fileWatcherType = "none"
-
-[logger]
-level = "info" 

web_app/app.py

@@ -1,30 +0,0 @@
-import streamlit as st
-import os
-import sys
-from pathlib import Path
-
-# 确保能找到项目模块
-sys.path.append(str(Path(__file__).parent))
-
-from pages import emotion_analyzer, chatbot  # 导入情绪分析页面和 Chatbot 页面
-
-def main():
-    st.set_page_config(
-        page_title="Audio Emotion Recognition System",
-        page_icon="🎵",
-        layout="wide"
-    )
-    
-    st.title("Audio Emotion Recognition System")
-    st.write("This is a web application for audio emotion recognition.")
-    
-    # 选择页面
-    page = st.sidebar.selectbox("Select a page", ["Emotion Analyzer", "Chatbot"])
-    
-    if page == "Emotion Analyzer":
-        emotion_analyzer.show()
-    elif page == "Chatbot":
-        chatbot.show_chatbot()
-
-if __name__ == "__main__":
-    main() 

web_app/components/__init__.py

@@ -1 +0,0 @@
-# 空文件，使components成为一个Python包 

web_app/components/audio_player.py

@@ -1,10 +0,0 @@
-import streamlit as st
-
-def play_audio(audio_file):
-    """
-    显示音频播放器组件
-    
-    Args:
-        audio_file: 上传的音频文件
-    """
-    st.audio(audio_file) 

web_app/components/debug_tools.py

@@ -1,56 +0,0 @@
-import streamlit as st
-import sys
-import os
-import platform
-from datetime import datetime
-
-class DebugTools:
-    @staticmethod
-    def show_debug_info():
-        """显示调试信息的可折叠部分"""
-        with st.expander("Debug Information", expanded=False):
-            # 系统信息
-            st.subheader("System Information")
-            st.text(f"System: {platform.system()} {platform.version()}")
-            st.text(f"Python Version: {sys.version}")
-            
-            # 内存使用
-            try:
-                import psutil
-                process = psutil.Process(os.getpid())
-                st.text(f"Memory Usage: {process.memory_info().rss / 1024 / 1024:.2f} MB")
-            except ImportError:
-                st.text("Memory Usage: Unable to get (requires psutil)")
-            
-            # GPU信息
-            try:
-                import torch
-                if torch.cuda.is_available():
-                    st.text(f"GPU: {torch.cuda.get_device_name(0)}")
-                    st.text(f"GPU Memory: {torch.cuda.memory_allocated(0)/1024/1024:.2f}MB / "
-                           f"{torch.cuda.memory_reserved(0)/1024/1024:.2f}MB")
-                else:
-                    st.text("GPU: Not Available")
-            except Exception as e:
-                st.text("GPU Information Retrieval Failed")
-    
-    @staticmethod
-    def log_error(error, context=None):
-        """记录错误信息"""
-        timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
-        error_msg = f"[{timestamp}] Error: {str(error)}"
-        if context:
-            error_msg += f"\nContext: {context}"
-        
-        st.error(error_msg)
-        # 可以添加日志文件记录
-        print(error_msg, file=sys.stderr)
-    
-    @staticmethod
-    def show_audio_info(audio_file):
-        """显示音频文件信息"""
-        if audio_file is not None:
-            st.write("Audio File Information:")
-            st.text(f"File Name: {audio_file.name}")
-            st.text(f"File Size: {audio_file.size/1024:.2f} KB")
-            st.text(f"File Type: {audio_file.type}") 

web_app/components/progress_bar.py

@@ -1,7 +0,0 @@
-import streamlit as st
-
-def show_progress():
-    """
-    返回一个streamlit进度条组件
-    """
-    return st.progress(0) 

web_app/components/visualizations.py

@@ -1,35 +0,0 @@
-import streamlit as st
-import plotly.graph_objects as go
-import plotly.express as px
-
-def plot_emotion_distribution(emotion_dict):
-    """
-    绘制情绪概率分布图（雷达图）
-    
-    Args:
-        emotion_dict: 包含情绪标签和对应概率的字典
-    """
-    emotions = list(emotion_dict.keys())
-    probabilities = [float(emotion_dict[emotion].strip('%')) / 100 for emotion in emotions]  # 转换为浮点数
-
-    # 创建雷达图
-    fig = go.Figure()
-
-    fig.add_trace(go.Scatterpolar(
-        r=probabilities + [probabilities[0]],  # 闭合图形
-        theta=emotions + [emotions[0]],  # 闭合图形
-        fill='toself',
-        name='Emotion Distribution'
-    ))
-
-    fig.update_layout(
-        title="Emotion Distribution",
-        polar=dict(
-            radialaxis=dict(
-                visible=True,
-                range=[0, 1]  # 设置范围
-            )),
-        showlegend=False
-    )
-
-    st.plotly_chart(fig, use_container_width=True) 

web_app/pages/__init__.py

@@ -1 +0,0 @@
-# 空文件，使pages成为一个Python包 

web_app/pages/chatbot.py

@@ -1,97 +0,0 @@
-import streamlit as st
-import torch
-import torchaudio
-import json
-from openai import AzureOpenAI
-from openai.types.beta.threads import Message
-from safetensors.torch import load_file
-from transformers import AutoTokenizer, Wav2Vec2Processor, BertModel, Wav2Vec2Model
-from huggingface_hub import hf_hub_download
-from dotenv import load_dotenv
-from utils import model_inference
-import os
-
-# 加载环境变量
-load_dotenv(".env")
-api_endpoint = os.getenv("AZURE_OPENAI_ENDPOINT")
-api_key = os.getenv("AZURE_OPENAI_API_KEY")
-api_version = os.getenv("AZURE_OPENAI_API_VERSION")
-api_deployment_name = os.getenv("AZURE_OPENAI_DEPLOYMENT")
-
-# 初始化 OpenAI 客户端
-client = AzureOpenAI(api_key=api_key, api_version=api_version, azure_endpoint=api_endpoint)
-
-# 设定 Chatbot 角色
-instruction = (
-    "You are a psychiatrist talking to a patient who may be depressed. "
-    "You'll receive their emotional state and conversation text. "
-    "Your goal is to help them open up and guide them to a positive path. "
-    "Be friendly, professional, empathetic, and supportive."
-)
-
-# 设定 Chatbot 线程和助手
-if "thread" not in st.session_state:
-    st.session_state.thread = client.beta.threads.create()
-
-if "assistant" not in st.session_state:
-    assistant_id = "asst_Sb1W9jVTeL1iyzu6N5MilgA1"
-    try:
-        st.session_state.assistant = client.beta.assistants.retrieve(assistant_id=assistant_id)
-    except:
-        st.session_state.assistant = client.beta.assistants.create(
-            name="Depression Chatbot",
-            instructions=instruction,
-            model=api_deployment_name,
-        )
-
-# 发送消息到 Azure Chatbot
-def send_message_to_chatbot(user_input, emotion):
-    chat_history = client.beta.threads.messages.list(thread_id=st.session_state.thread.id)
-    messages = [{"role": msg.role, "content": msg.content} for msg in chat_history]
-    
-    messages.append({"role": "user", "content": f"Emotion: {emotion}. {user_input}"})
-
-    client.beta.threads.messages.create(
-        thread_id=st.session_state.thread.id,
-        role="user",
-        content=f"Emotion: {emotion}. {user_input}",
-    )
-
-    run = client.beta.threads.runs.create(
-        thread_id=st.session_state.thread.id,
-        assistant_id=st.session_state.assistant.id,
-    )
-
-    while run.status in ["queued", "in_progress"]:
-        run = client.beta.threads.runs.retrieve(run.id)
-
-    response_messages = client.beta.threads.messages.list(thread_id=st.session_state.thread.id)
-    return response_messages[-1].content if response_messages else "No response."
-
-# Streamlit 界面
-st.title("🧠 AI Depression Chatbot")
-
-if "chat_history" not in st.session_state:
-    st.session_state.chat_history = []
-
-# 用户输入
-user_input = st.text_input("Enter your message:")
-audio_file = st.file_uploader("Upload audio file", type=["wav", "mp3"])
-
-if st.button("Send"):
-    if user_input or audio_file:
-        emotion_probabilities = model_inference.predict_emotion(user_input, audio_file)
-        dominant_emotion = max(emotion_probabilities, key=emotion_probabilities.get)
-
-        chatbot_response = send_message_to_chatbot(user_input, dominant_emotion)
-        
-        # 保存聊天记录
-        st.session_state.chat_history.append({"role": "user", "content": user_input})
-        st.session_state.chat_history.append({"role": "assistant", "content": chatbot_response})
-
-        # 显示聊天记录
-        for chat in st.session_state.chat_history:
-            st.write(f"**{chat['role'].capitalize()}**: {chat['content']}")
-
-    else:
-        st.warning("Please enter a message or upload an audio file.")

web_app/pages/emotion_analyzer.py

@@ -1,80 +0,0 @@
-import streamlit as st
-from components.visualizations import plot_emotion_distribution
-from utils import model_inference
-from components.audio_player import play_audio
-from components.debug_tools import DebugTools
-import json
-import os
-
-def show_history():
-    """显示历史记录"""
-    history_file = "history.json"
-    
-    if os.path.exists(history_file):
-        with open(history_file, 'r') as f:
-            history = json.load(f)
-        
-        if history:
-            st.subheader("History")
-            for record in history:
-                st.write(f"Audio File: {record['audio_file']}")
-                st.write(f"Transcript: {record['transcript']}")
-                st.write(f"Emotions: {record['emotions']}")
-                st.write(f"Probabilities: {record['probabilities']}")
-                st.write("---")
-        else:
-            st.write("No history records.")
-    else:
-        st.write("No history file.")
-
-def show():
-    st.header("Emotion Analyzer")
-    
-    # 显示历史记录
-    show_history()
-    
-    # 初始化调试工具
-    debug = DebugTools()
-    
-    # 显示系统调试信息
-    debug.show_debug_info()
-    
-    # 文件上传
-    audio_file = st.file_uploader("Upload audio file", type=['wav', 'mp3'])
-    text_input = st.text_input("Enter text input")
-
-    if audio_file is not None and text_input:
-        # 显示音频文件信息
-        debug.show_audio_info(audio_file)
-        
-        # 使用audio_player组件
-        play_audio(audio_file)
-        
-        if st.button("Analyse Your Emotion!😊"):
-            # 显示进度条
-            progress_bar = st.progress(0)
-            
-            try:
-                # 直接使用用户输入的文本作为转写
-                transcript = text_input
-                st.write("Audio transcript:", transcript)
-                
-                # 2. 情绪分析
-                with st.spinner("Analysing emotion..."):
-                    progress_bar.progress(30)
-                    emotions = model_inference.predict_emotion(text_input, audio_file)
-                
-                # 3. 显示结果
-                progress_bar.progress(30)
-                
-                # 显示预测结果
-                st.success(f"Predict: {emotions}")
-                
-                # 显示情绪概率分布图
-                plot_emotion_distribution(emotions)
-                
-                # 保存历史记录
-                model_inference.save_history(audio_file, transcript, emotions, None)  # 这里可以根据需要调整
-                
-            except Exception as e:
-                debug.log_error(e, context=f"Processing file: {audio_file.name}")

web_app/requirements.txt

@@ -1,14 +0,0 @@
-streamlit>=1.0.0
-plotly>=5.0.0
-librosa>=0.9.0
-scipy>=1.7.0
-numpy>=1.21.0
-torch>=1.9.0
-transformers>=4.11.0
-soundfile>=0.10.3
-psutil>=5.8.0
-huggingface-hub>=0.0.12
-safetensors>=0.0.3
-torchaudio>=0.9.0
-openai
-dotenv

web_app/utils/__init__.py

@@ -1 +0,0 @@
-# 空文件，使utils成为一个Python包 

web_app/utils/audio_processing.py

@@ -1,31 +0,0 @@
-import librosa
-import numpy as np
-from scipy.signal import butter, filtfilt
-
-class AudioProcessor:
-    @staticmethod
-    def load_and_process_audio(file_path, target_sr=16000):
-        """加载并处理音频文件"""
-        # 加载音频文件
-        audio_data, sr = librosa.load(file_path, sr=target_sr)
-        
-        # 归一化音频数据
-        audio_data = librosa.util.normalize(audio_data)
-        
-        return audio_data
-
-    @staticmethod
-    def resample(audio_data, orig_sr, target_sr=16000):
-        """重采样音频"""
-        return librosa.resample(audio_data, orig_sr=orig_sr, target_sr=target_sr)
-    
-    @staticmethod
-    def denoise(audio_data, sr):
-        """音频降噪"""
-        # 实现降噪逻辑
-        return audio_data
-    
-    @staticmethod
-    def normalize(audio_data):
-        """音频归一化"""
-        return librosa.util.normalize(audio_data) 

web_app/utils/logger.py

@@ -1,37 +0,0 @@
-import logging
-import os
-from datetime import datetime
-
-class Logger:
-    def __init__(self):
-        # 创建logs目录
-        log_dir = "logs"
-        if not os.path.exists(log_dir):
-            os.makedirs(log_dir)
-            
-        # 设置日志文件名
-        log_file = os.path.join(
-            log_dir, 
-            f"app_{datetime.now().strftime('%Y%m%d_%H%M%S')}.log"
-        )
-        
-        # 配置日志
-        logging.basicConfig(
-            level=logging.INFO,
-            format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
-            handlers=[
-                logging.FileHandler(log_file),
-                logging.StreamHandler()
-            ]
-        )
-        
-        self.logger = logging.getLogger('emotion_recognition')
-    
-    def info(self, message):
-        self.logger.info(message)
-    
-    def error(self, message, exc_info=True):
-        self.logger.error(message, exc_info=exc_info)
-    
-    def debug(self, message):
-        self.logger.debug(message) 

web_app/utils/model_inference.py

@@ -1,89 +0,0 @@
-import torch
-import json
-import os
-from transformers import AutoTokenizer, BertModel, Wav2Vec2Model
-from utils.audio_processing import AudioProcessor
-import torchaudio
-import torch.nn.functional as F
-from huggingface_hub import hf_hub_download
-from safetensors.torch import load_file
-
-# 下载模型
-model_path = hf_hub_download(repo_id="liloge/Group7_model_test", filename="model.safetensors")
-
-class MultimodalClassifier(torch.nn.Module):
-    def __init__(self):
-        super(MultimodalClassifier, self).__init__()
-        self.bert = BertModel.from_pretrained("bert-base-uncased")
-        self.wav2vec2 = Wav2Vec2Model.from_pretrained("facebook/wav2vec2-base-960h")
-        self.classifier = torch.nn.Sequential(
-            torch.nn.Linear(self.bert.config.hidden_size + self.wav2vec2.config.hidden_size, 256),
-            torch.nn.ReLU(),
-            torch.nn.Dropout(0.7),
-            torch.nn.Linear(256, 7)  # 7分类任务
-        )
-
-    def forward(self, text_input, audio_input):
-        text_outputs = self.bert(**text_input, output_hidden_states=True)
-        text_features = text_outputs.hidden_states[-1][:, 0, :]  # [CLS] token
-        audio_outputs = self.wav2vec2(audio_input, output_hidden_states=True)
-        audio_features = audio_outputs.hidden_states[-1][:, 0, :]
-        combined_features = torch.cat((text_features, audio_features), dim=-1)
-        logits = self.classifier(combined_features)
-        return logits
-
-# 加载模型
-device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-model = MultimodalClassifier().to(device)
-
-# 加载 SafeTensors 权重
-state_dict = load_file(model_path)
-model.load_state_dict(state_dict)  
-model.eval()  # 设置为评估模式
-
-tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
-
-def preprocess_text(text):
-    return tokenizer(text, return_tensors="pt", truncation=True, padding=True, max_length=128).to(device)
-
-def preprocess_audio(audio_path):
-    waveform, sample_rate = torchaudio.load(audio_path)
-    waveform = torchaudio.transforms.Resample(orig_freq=sample_rate, new_freq=16000)(waveform)
-    return waveform.to(device)
-
-labels = ["Neutral", "Happy", "Sad", "Angry", "Fearful", "Disgusted", "Surprised"]
-
-def predict_emotion(text, audio):
-    text_inputs = preprocess_text(text)
-    audio_inputs = preprocess_audio(audio)
-
-    with torch.no_grad():
-        output = model(text_inputs, audio_inputs)  # (1, 7) logits
-        probabilities = F.softmax(output, dim=1).squeeze().tolist()  # 归一化为概率
-
-    return {labels[i]: f"{probabilities[i]*100:.2f}%" for i in range(len(labels))}
-
-def generate_transcript(audio_file):
-    """生成音频的文字转写"""
-    return audio_file.name  # 直接返回音频文件的名称
-
-def save_history(audio_file, transcript, emotions, probabilities):
-    """保存分析历史记录到文件"""
-    history_file = "history.json"
-    
-    if not os.path.exists(history_file):
-        with open(history_file, 'w') as f:
-            json.dump([], f)
-    
-    with open(history_file, 'r') as f:
-        history = json.load(f)
-    
-    history.append({
-        "audio_file": audio_file.name,
-        "transcript": transcript,
-        "emotions": emotions,
-        "probabilities": probabilities
-    })
-    
-    with open(history_file, 'w') as f:
-        json.dump(history, f, indent=4)