Update app.py

app.py

@@ -18,7 +18,6 @@ logging.getLogger('matplotlib').setLevel(logging.WARNING)
 
 limitation = os.getenv("SYSTEM") == "spaces"  # limit audio length in huggingface spaces
 
-
 audio_postprocess_ori = gr.Audio.postprocess
 
 
@@ -43,20 +42,17 @@ def create_vc_fn(model, spk):
         audio = (audio / np.iinfo(audio.dtype).max).astype(np.float32)
         if len(audio.shape) > 1:
             audio = librosa.to_mono(audio.transpose(1, 0))
-        if sampling_rate != 16000:
-            audio = librosa.resample(audio, orig_sr=sampling_rate, target_sr=16000)
 
         raw_audio_path = io.BytesIO()
         soundfile.write(raw_audio_path, audio, sampling_rate, format="wav")
         raw_audio_path.seek(0)
-        # out_audio, _, _ = model.infer(raw_path=raw_audio_path,
-        #                               speaker=spk,
-        #                               tran=vc_transform,
-        #                               f0_predictor=f0p,
-        #                               auto_predict_f0=auto_f0)
+        out_audio, _, _ = model.infer(raw_path=raw_audio_path,
+                                      speaker=spk,
+                                      tran=vc_transform,
+                                      f0_predictor=f0p,
+                                      auto_predict_f0=auto_f0)
 
-        # return "Success", (44100, out_audio.cpu().numpy())
-        return 'test', (44100, audio)
+        return 44100, out_audio.cpu().numpy()
 
     return vc_fn
 
@@ -91,14 +87,13 @@ if __name__ == '__main__':
                             )
                         with gr.Column():
                             vc_input = gr.Audio(label="输入干声" + ' (小于 20 秒)' if limitation else '')
-                            vc_transform = gr.Number(label="音高调整(支持正负半音，12为一个八度)", value=0)
-                            auto_f0 = gr.Checkbox(label="自动音高预测(正常说话可选)", value=False)
-                            f0_predictor = gr.Radio(label="f0预测器(对电音大小有影响，可以都试试)",
+                            vc_transform = gr.Number(label="音高调整 (支持正负半音，12为一个八度)", value=0)
+                            auto_f0 = gr.Checkbox(label="自动音高预测 (正常说话可选)", value=False)
+                            f0_predictor = gr.Radio(label="f0预测器 (对电音有影响)",
                                                     choices=['crepe', 'harvest', 'dio', 'pm'], value='crepe')
                             vc_submit = gr.Button("生成", variant="primary")
 
                         with gr.Column():
-                            vc_output1 = gr.Textbox(label="输出信息")
-                            vc_output2 = gr.Audio(label="输出音频")
-                vc_submit.click(vc_fn, [vc_input, vc_transform, auto_f0, f0_predictor], [vc_output1, vc_output2])
+                            vc_output = gr.Audio(label="输出音频")
+                vc_submit.click(vc_fn, [vc_input, vc_transform, auto_f0, f0_predictor], vc_output)
         app.queue(concurrency_count=1, api_open=args.api).launch(share=args.share)