add tts 2D2

.gitignore

@@ -1 +1,12 @@
-__pycache__/
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.env
+.venv
+venv*
+env/
+ENV/
+*.wav
+test_*.py
+.DS_Store

app.py

@@ -417,24 +417,30 @@ def generate_tts_with_pauses(text, silence_duration_ms=500):
 
     # 逐句生成音频
     audio_segments = []
+    sampling_rate = None
+
     for sentence in sentences:
         if sentence:
             result = tts(sentence)
             audio_segments.append(result["audio"])
+            if sampling_rate is None:
+                sampling_rate = result["sampling_rate"]
 
     # 拼接音频，在之间添加静音
-    silence_samples = int(result["sampling_rate"] * silence_duration_ms / 1000)
+    silence_samples = int(sampling_rate * silence_duration_ms / 1000)
     silence = np.zeros(silence_samples, dtype=np.float32)
 
     combined_audio = []
     for i, segment in enumerate(audio_segments):
+        # 统一转换为 1D numpy 数组，避免不同环境返回的格式不一致（如 torch Tensor / numpy 多维数组）
+        segment = np.asarray(segment).flatten()
         combined_audio.append(segment)
         if i < len(audio_segments) - 1:  # 最后一段后面不加静音
             combined_audio.append(silence)
 
     final_audio = np.concatenate(combined_audio)
 
-    return {"audio": final_audio, "sampling_rate": result["sampling_rate"]}
+    return {"audio": final_audio, "sampling_rate": sampling_rate}
 
 
 @app.post("/upload/text")