fix

app.py

@@ -14,6 +14,7 @@ from audio2numpy import open_audio
 from spleeter.separator import Separator
 from spleeter.audio.adapter import AudioAdapter
 from pydub import AudioSegment
+import scipy.io.wavfile
 
 import logging
 
@@ -168,19 +169,29 @@ def svc_change(argswave, argsspk):
     return out_audio
 
 
+def np_to_audio_segment(fp_arr):
+    wav_io = io.BytesIO()
+    scipy.io.wavfile.write(wav_io, 16000, fp_arr)
+    wav_io.seek(0)
+    sound = AudioSegment.from_wav(wav_io)
+    return sound
+
+
 def svc_main(sid, input_audio):
     if input_audio is None:
         return "You need to upload an audio", None
     sampling_rate, audio = input_audio
+    input_audio_tmp_file = 'origin.wav'
     #
-    prediction = separator.separate(audio)
-    vocals, accompaniment = prediction["vocals"], prediction["accompaniment"]
+    # prediction = separator.separate(audio)
+    # vocals, accompaniment = prediction["vocals"], prediction["accompaniment"]
+    soundfile.write(input_audio_tmp_file, audio, sampling_rate, format="wav")
+    separator.separate_to_file(input_audio_tmp_file, '')
+
+    vocals_filepath = os.path.join(os.path.splitext(input_audio_tmp_file)[0], 'vocals.wav')
+    accompaniment_filepath = os.path.join(os.path.splitext(input_audio_tmp_file)[0], 'accompaniment.wav')
 
-    # 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)
-    soundfile.write('vocals_111.wav', vocals, sampling_rate, format="wav")
-    soundfile.write('accompaniment_111.wav', accompaniment, sampling_rate, format="wav")
+    vocals, sampling_rate = soundfile.read(vocals_filepath)
 
     vocals = (vocals / np.iinfo(vocals.dtype).max).astype(np.float32)
     if len(vocals.shape) > 1:
@@ -193,21 +204,11 @@ def svc_main(sid, input_audio):
     soundfile.write(wav_path, vocals, 16000, format="wav")
 
     out_vocals = svc_change(wav_path, f"configs/singers/singer00{sid}.npy")
+    out_vocals_filepath = os.path.join(os.path.splitext(input_audio_tmp_file)[0], 'out_vocals.wav')
+    soundfile.write(out_vocals_filepath, out_vocals, 48000, format="wav")
 
-    soundfile.write('out_vocals_111.wav', out_vocals, 48000, format="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
-    )
+    sound1 = AudioSegment.from_file(out_vocals_filepath)
+    sound2 = AudioSegment.from_file(accompaniment_filepath)
 
     played_togther = sound1.overlay(sound2)