Update app.py

app.py

@@ -330,7 +330,7 @@ def voice_conversion(source, target, diffusion_steps, length_adjust, inference_c
             processed_frames += vc_target.size(2) - overlap_frame_len
             output_wave = (output_wave * 32768.0).astype(np.int16)
             mp3_bytes = AudioSegment(
-                output_wave.tobytes(), frame_rate=sr,
+             output_wave.tobytes(), frame_rate=sr,
                 sample_width=output_wave.dtype.itemsize, channels=1
             ).export(format="mp3", bitrate=bitrate).read()
             yield mp3_bytes, None
@@ -345,7 +345,7 @@ if __name__ == "__main__":
     inputs = [
         gr.Audio(type="filepath", label="Source Audio / 源音频"),
         gr.Audio(type="filepath", label="Reference Audio / 参考音频"),
-        gr.Slider(minimum=1, maximum=200, value=25, step=1, label="Diffusion Steps / 扩散步数", info="25 by default, 50~100 for best quality / 默认为 25，50~100 为最佳质量"),
+        gr.Slider(minimum=1, maximum=200, value=4, step=1, label="Diffusion Steps / 扩散步数", info="25 by default, 50~100 for best quality / 默认为 25，50~100 为最佳质量"),
         gr.Slider(minimum=0.5, maximum=2.0, step=0.1, value=1.0, label="Length Adjust / 长度调整", info="<1.0 for speed-up speech, >1.0 for slow-down speech / <1.0 加速语速，>1.0 减慢语速"),
         gr.Slider(minimum=0.0, maximum=1.0, step=0.1, value=0.7, label="Inference CFG Rate", info="has subtle influence / 有微小影响"),
         gr.Checkbox(label="Use F0 conditioned model / 启用F0输入", value=False, info="Must set to true for singing voice conversion / 歌声转换时必须勾选"),