fix

app.py

@@ -177,8 +177,10 @@ def svc_main(sid, input_audio):
     vocals, accompaniment = prediction["vocals"], prediction["accompaniment"]
 
     # 歌词和伴奏是否有问题
-    audio_loader.save('vocals_111.wav', vocals, sampling_rate)
-    audio_loader.save('accompaniment_111.wav', accompaniment, sampling_rate)
+    # audio_loader.save('vocals_111.wav', vocals, sampling_rate)
+    # audio_loader.save('accompaniment_111.wav', accompaniment, sampling_rate)
+    soundfile.write('vocals_111.wav', vocals, sampling_rate, format="wav")
+    soundfile.write('accompaniment_111.wav', accompaniment, sampling_rate, format="wav")
 
     vocals = (vocals / np.iinfo(vocals.dtype).max).astype(np.float32)
     if len(vocals.shape) > 1:
@@ -192,10 +194,20 @@ def svc_main(sid, input_audio):
 
     out_vocals = svc_change(wav_path, f"configs/singers/singer00{sid}.npy")
 
-    audio_loader.save('out_vocals_111.wav', out_vocals, 48000)
+    soundfile.write('out_vocals_111.wav', out_vocals, 48000, format="wav")
 
-    sound1 = AudioSegment.from_file("out_vocals_111.wav")
-    sound2 = AudioSegment.from_file("accompaniment_111.wav")
+    sound1 = AudioSegment.from_file(
+        out_vocals[:, 0].tobytes(),
+        frame_rate=sampling_rate,
+        sample_width=out_vocals[:, 0].dtype.itemsize,
+        channels=1
+    )
+    sound2 = AudioSegment(
+        accompaniment[:, 0].tobytes(),
+        frame_rate=sampling_rate,
+        sample_width=accompaniment[:, 0].dtype.itemsize,
+        channels=1
+    )
 
     played_togther = sound1.overlay(sound2)