Upload infertest.py

infertest.py

@@ -26,6 +26,7 @@ tmp = os.path.join(now_dir, "TEMP")
 shutil.rmtree(tmp, ignore_errors=True)
 os.makedirs(tmp, exist_ok=True)
 os.environ["TEMP"] = tmp
+split_model=htdemucs
 from lib.infer_pack.models import (
     SynthesizerTrnMs256NSFsid,
     SynthesizerTrnMs256NSFsid_nono,
@@ -119,8 +120,17 @@ def api_convert_voice():
     input_audio_path = os.path.join(tmp, f"{spk_id}_input_audio.{filename.split('.')[-1]}")
     file.save(input_audio_path)
     
-    output_path = convert_voice(spk_id, input_audio_path, voice_transform)
+    #split audio
+    cut_vocal_and_inst(input_audio_path)
+    print("audio splitting performed")
+    vocal_path = f"output/{split_model}/audio/vocals.wav"
+    inst = f"output/{split_model}/audio/no_vocals.wav"
+    
+    output_path = convert_voice(spk_id, vocal_path, voice_transform)
+    output_path= combine_vocal_and_inst(output_path,inst)
     print(output_path)
+    
+    
     if os.path.exists(output_path):
         return send_file(output_path, as_attachment=True)
     else:
@@ -441,34 +451,47 @@ def cut_vocal_and_inst_yt(split_model):
     logs.append("Audio splitting complete.")
     yield "\n".join(logs), vocal, inst, vocal
 
-def cut_vocal_and_inst(split_model, audio_data):
+def cut_vocal_and_inst(audio_path):
     logs = []
     vocal_path = "output/result/audio.wav"
     os.makedirs("output/result", exist_ok=True)
-    wavfile.write(vocal_path, audio_data[0], audio_data[1])
-    logs.append("Starting the audio splitting process...")
-    yield "\n".join(logs), None, None
+    #wavfile.write(vocal_path, audio_data[0], audio_data[1])
+    #logs.append("Starting the audio splitting process...")
+    #yield "\n".join(logs), None, None
     command = f"demucs --two-stems=vocals -n {split_model} {vocal_path} -o output"
     result = subprocess.Popen(command.split(), stdout=subprocess.PIPE, text=True)
-    for line in result.stdout:
-        logs.append(line)
-        yield "\n".join(logs), None, None
+    #for line in result.stdout:
+    #    logs.append(line)
+    #    yield "\n".join(logs), None, None
     print(result.stdout)
     vocal = f"output/{split_model}/audio/vocals.wav"
     inst = f"output/{split_model}/audio/no_vocals.wav"
-    logs.append("Audio splitting complete.")
-    yield "\n".join(logs), vocal, inst
-    
-def combine_vocal_and_inst(audio_data, vocal_volume, inst_volume, split_model):
+    #logs.append("Audio splitting complete.")
+
+
+def combine_vocal_and_inst(vocal_path, inst_path):
+    vocal_volume=1
+    inst_volume=1
     os.makedirs("output/result", exist_ok=True)
-    vocal_path = "output/result/output.wav"
+    # Assuming vocal_path and inst_path are now directly passed as arguments
     output_path = "output/result/combine.mp3"
-    inst_path = f"output/{split_model}/audio/no_vocals.wav"
-    wavfile.write(vocal_path, audio_data[0], audio_data[1])
-    command =  f'ffmpeg -y -i {inst_path} -i {vocal_path} -filter_complex [0:a]volume={inst_volume}[i];[1:a]volume={vocal_volume}[v];[i][v]amix=inputs=2:duration=longest[a] -map [a] -b:a 320k -c:a libmp3lame {output_path}'
-    result = subprocess.run(command.split(), stdout=subprocess.PIPE)
-    print(result.stdout.decode())
+    command = f'ffmpeg -y -i "{inst_path}" -i "{vocal_path}" -filter_complex [0:a]volume={inst_volume}[i];[1:a]volume={vocal_volume}[v];[i][v]amix=inputs=2:duration=longest[a] -map [a] -b:a 320k -c:a libmp3lame "{output_path}"'
+    result = subprocess.run(command.split(), stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+    if result.stderr:
+        print("Error:", result.stderr.decode())
+    else:
+        print("Success:", result.stdout.decode())
     return output_path
+    
+#def combine_vocal_and_inst(audio_data, vocal_volume, inst_volume):
+#    os.makedirs("output/result", exist_ok=True)
+ ##  output_path = "output/result/combine.mp3"
+   # inst_path = f"output/{split_model}/audio/no_vocals.wav"
+    #wavfile.write(vocal_path, audio_data[0], audio_data[1])
+    #command =  f'ffmpeg -y -i {inst_path} -i {vocal_path} -filter_complex [0:a]volume={inst_volume}[i];[1:a]volume={vocal_volume}[v];[i][v]amix=inputs=2:duration=longest[a] -map [a] -b:a 320k -c:a libmp3lame {output_path}'
+    #result = subprocess.run(command.split(), stdout=subprocess.PIPE)
+    #print(result.stdout.decode())
+    #return output_path
 
 def download_and_extract_models(urls):
     logs = []