Update app.py

app.py

@@ -8,9 +8,6 @@ import tempfile
 import shutil
 import requests
 from pathlib import Path
-global ckpt_temp_file
-global audio_temp_file
-global config_temp_file
 ###################################################
 from utils.hparams import hparams
 from preprocessing.data_gen_utils import get_pitch_parselmouth,get_pitch_crepe
@@ -57,7 +54,22 @@ def vc_fn(sid, audio_record, audio_upload, tran, pndm_speedup=20):
     f0_tst, f0_pred, audio = run_clip(svc_model, file_path=audio_path, key=key, acc=pndm_speedup, use_crepe=True, use_pe=True, thre=0.05,
                                             use_gt_mel=False, add_noise_step=500, project_name=project_name, out_path=wav_gen)
 
-    return "Success", (hparams['audio_sample_rate'], audio)
+    f0_gen,_=get_pitch_parselmouth(*svc_model.vocoder.wav2spec(wav_gen),hparams)
+    f0_tst[f0_tst==0]=np.nan#ground truth f0
+    f0_pred[f0_pred==0]=np.nan#f0 pe predicted
+    f0_gen[f0_gen==0]=np.nan#f0 generated
+    fig=plt.figure(figsize=[15,5])
+    plt.plot(np.arange(0,len(f0_tst)),f0_tst,color='black')
+    plt.plot(np.arange(0,len(f0_pred)),f0_pred,color='orange')
+    plt.plot(np.arange(0,len(f0_gen)),f0_gen,color='red')
+    plt.axhline(librosa.note_to_hz('C4'),ls=":",c="blue")
+    plt.axhline(librosa.note_to_hz('G4'),ls=":",c="green")
+    plt.axhline(librosa.note_to_hz('C5'),ls=":",c="orange")
+    plt.axhline(librosa.note_to_hz('F#5'),ls=":",c="red")
+    #plt.axhline(librosa.note_to_hz('A#5'),ls=":",c="black")
+    plt.savefig('./temp.png')
+
+    return "Success", (hparams['audio_sample_rate'], audio), gr.Image.update("temp.png")
 
 
 app = gr.Blocks()
@@ -88,9 +100,9 @@ with app:
             gr.Markdown(value="""f0曲线可以直观的显示跑调情况，蓝色为输入音高，橙色为合成音频的音高
                         若**只看见橙色**，说明蓝色曲线被覆盖，转换效果较好
                         """)
-            # f0_image = gr.Image(label="f0曲线")
+            f0_image = gr.Image(label="f0曲线")
         vc_submit.click(vc_fn, [speaker_id, record_input, upload_input, vc_transform, vc_speedup],
-                        [out_message, out_audio])
+                        [out_message, out_audio, f0_image])
         with gr.TabItem("使用说明"):
             gr.Markdown(value="""
                         0、合集：https://github.com/IceKyrin/sovits_guide/blob/main/README.md