Update app_colab.py

app_colab.py

@@ -5,6 +5,10 @@ import torch
 
 device = 'cuda' if torch.cuda.is_available() else 'cpu'
 
+import zipfile
+with zipfile.ZipFile("speech_XXL_cond.zip", 'r') as zip_ref:
+    zip_ref.extractall("Phoneme_Hallucinator_v2/exp")
+
 headers = {
     "user-agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/121.0.0.0 Safari/537.36"
 }
@@ -75,7 +79,6 @@ for name in os.listdir(weight_uvr5_root):
         uvr5_names.append(name.replace(".pth", ""))
 
 func = AudioPre
-
 pre_fun_hp2 = func(
   agg=int(10),
   model_path=os.path.join(weight_uvr5_root, "UVR-HP2.pth"),
@@ -214,7 +217,7 @@ def youtube_downloader_100s(
     return f"./output/{split_model}/{filename}/vocal_{filename}.wav_10.wav", f"./output/{split_model}/{filename}/instrument_{filename}.wav_10.wav"
 
 
-def convert(start_time, song_name_src, song_name_ref, check_song, key_shift, vocal_vol, inst_vol):
+def convert(start_time, song_name_src, song_name_ref, check_song, auto_key, key_shift, vocal_vol, inst_vol):
   split_model = "UVR-HP5"
   song_name_ref = song_name_ref.strip().replace(" ", "")
   video_identifier = search_bilibili(song_name_ref)
@@ -243,10 +246,17 @@ def convert(start_time, song_name_src, song_name_ref, check_song, key_shift, voc
     os.remove("output_svc/NeuCoSVCv2.wav")
 
   if check_song == True:
-    os.system(f"python inference.py --src_wav_path audio_src.wav --ref_wav_path voiced_audio.wav --key_shift {key_shift}")
+      if auto_key == True:
+          os.system(f"python inference.py --src_wav_path audio_src.wav --ref_wav_path voiced_audio.wav")
+      else:
+          os.system(f"python inference.py --src_wav_path audio_src.wav --ref_wav_path voiced_audio.wav --key_shift {key_shift}")
+ 
   else:
-    os.system(f"python inference.py --src_wav_path audio_src.wav --ref_wav_path voiced_audio.wav --key_shift {key_shift} --speech_enroll")
-
+      if auto_key == True:
+          os.system(f"python inference.py --src_wav_path audio_src.wav --ref_wav_path voiced_audio.wav --speech_enroll")
+      else:
+          os.system(f"python inference.py --src_wav_path audio_src.wav --ref_wav_path voiced_audio.wav --key_shift {key_shift} --speech_enroll")
+          
   audio_vocal = AudioSegment.from_file("output_svc/NeuCoSVCv2.wav", format="wav")
 
   # Load the second audio file
@@ -271,7 +281,7 @@ app = gr.Blocks()
 with app:
   gr.Markdown("# <center>🥳💕🎶 NeuCoSVC v2 AI歌手全明星，无需训练、一键翻唱、重磅更新！</center>")
   gr.Markdown("## <center>🌟 只需 1 个歌曲名，一键翻唱任意歌手的任意歌曲，支持说话语音翻唱，随时随地，听你想听！</center>")
-  gr.Markdown("### <center>🌊 NeuCoSVC v2 先享版 Powered by Tencent ARC Lab & Tsinghua University 💕</center>")
+  gr.Markdown("### <center>🌊 [NeuCoSVC v2](https://github.com/thuhcsi/NeuCoSVC) 先享版 Powered by Tencent ARC Lab & Tsinghua University 💕</center>")
   with gr.Row():
     with gr.Column():
       with gr.Row():
@@ -280,15 +290,16 @@ with app:
       with gr.Row():
         inp0 = gr.Number(value=0, label="起始时间 (秒)", info="此程序将自动从起始时间开始提取45秒的翻唱歌曲")
         inp3 = gr.Checkbox(label="参考音频是否为歌曲演唱，默认为是", info="如果参考音频为正常说话语音，请取消打勾", value=True)
-        inp4 = gr.Slider(minimum=-12, maximum=12, value=0, step=1, label="歌曲人声升降调", info="默认为0，+2为升高2个key，以此类推")
+        inp4 = gr.Checkbox(label="是否自动预测歌曲人声升降调，默认为是", info="如果需要手动调节歌曲人声升降调，请取消打勾", value=True)
+        inp5 = gr.Slider(minimum=-12, maximum=12, value=0, step=1, label="歌曲人声升降调", info="默认为0，+2为升高2个key，以此类推")
       with gr.Row():
-        inp5 = gr.Slider(minimum=-3, maximum=3, value=0, step=1, label="调节人声音量，默认为0")
-        inp6 = gr.Slider(minimum=-3, maximum=3, value=0, step=1, label="调节伴奏音量，默认为0")
+        inp6 = gr.Slider(minimum=-3, maximum=3, value=0, step=1, label="调节人声音量，默认为0")
+        inp7 = gr.Slider(minimum=-3, maximum=3, value=0, step=1, label="调节伴奏音量，默认为0")
       btn = gr.Button("一键开启AI翻唱之旅吧💕", variant="primary")
     with gr.Column():
-      out = gr.Audio(label="AI歌手为您倾情演唱的歌曲", type="filepath", interactive=True)
+      out = gr.Audio(label="AI歌手为您倾情演唱的歌曲", type="filepath", interactive=False)
 
-  btn.click(convert, [inp0, inp1, inp2, inp3, inp4, inp5, inp6], out)
+  btn.click(convert, [inp0, inp1, inp2, inp3, inp4, inp5, inp6, inp7], out)
 
   gr.Markdown("### <center>注意❗：请不要生成会对个人以及组织造成侵害的内容，此程序仅供科研、学习及个人娱乐使用。</center>")
   gr.HTML('''