requirements.txt

gradio==3.37.0
librosa
numpy
scikit-learn
deepface
opencv-python
joblib

app.py

@@ -0,0 +1,48 @@
+import gradio as gr
+import librosa
+import numpy as np
+from sklearn.ensemble import RandomForestClassifier
+import joblib
+from deepface import DeepFace
+
+# —— 假设你的语音模型已经训练好并存为 voice_model.joblib ——  
+voice_clf = joblib.load("voice_model.joblib")
+
+# 文字分析
+def analyze_text(text):
+    if any(w in text for w in ["開心","快樂"]): return "happy"
+    if any(w in text for w in ["生氣","憤怒"]): return "angry"
+    if any(w in text for w in ["傷心","難過","哭"]): return "sad"
+    if any(w in text for w in ["驚","意外"]): return "surprise"
+    if any(w in text for w in ["怕","恐懼"]): return "fear"
+    return "neutral"
+
+# 語音分析
+def analyze_audio(path):
+    y_audio, sr = librosa.load(path, sr=None)
+    mfccs = np.mean(librosa.feature.mfcc(y=y_audio, sr=sr, n_mfcc=13).T, axis=0).reshape(1, -1)
+    return voice_clf.predict(mfccs)[0]
+
+# 臉部分析
+def analyze_face(img):
+    res = DeepFace.analyze(img, actions=['emotion'], enforce_detection=False)
+    return res[0]['dominant_emotion']
+
+# 定義 Gradio 介面
+with gr.Blocks() as demo:
+    gr.Markdown("# 多模態情緒分析")
+    with gr.Tab("📝 文字"):
+        t = gr.Textbox(placeholder="輸入中文…")
+        bt = gr.Button("分析文字")
+        out_t = gr.Textbox()
+        bt.click(analyze_text, inputs=t, outputs=out_t)
+    with gr.Tab("🎤 語音"):
+        a = gr.Audio(type="filepath")
+        ba = gr.Button("分析語音")
+        out_a = gr.Textbox()
+        ba.click(analyze_audio, inputs=a, outputs=out_a)
+    with gr.Tab("📷 臉部"):
+        im = gr.Image(source="webcam")
+        bi = gr.Button("分析表情")
+        out_i = gr.Textbox()
+        bi.click(analyze_face, inputs=im, outputs=out_i)