upd req

app.py

@@ -30,8 +30,6 @@ def download_audio(url: str, save_path: str):
         for chunk in response.iter_content(chunk_size=8192):
             file.write(chunk)
 
-    print(f"{url} has been downloded to {save_path}")
-
 
 def is_url(s: str):
     try:
@@ -43,18 +41,13 @@ def is_url(s: str):
 
 
 def audio2midi(audio_path: str, cache_dir: str):
-    # Load audio
-    audio, _ = load_audio(audio_path, sr=sample_rate, mono=True)
-    print("load_audio OK")
+    audio, _ = load_audio(audio_path, sr=sample_rate)
     transcriptor = PianoTranscription(
         device="cuda" if torch.cuda.is_available() else "cpu",
         checkpoint_path=WEIGHTS_PATH,
     )
-    print("PianoTranscription OK")
     midi_path = f"{cache_dir}/output.mid"
-    # midi_path = audio_path.replace(audio_path.split(".")[-1], "mid")
     transcriptor.transcribe(audio, midi_path)
-    print("transcribe OK")
     return midi_path, os.path.basename(audio_path).split(".")[-2].capitalize()
 
 
@@ -76,7 +69,6 @@ def get_1st_int(input_string: str):
     match = re.search(r"\d+", input_string)
     if match:
         return str(int(match.group()))
-
     else:
         return ""
 
@@ -121,15 +113,10 @@ def url_infer(song: str, cache_dir="./__pycache__/mode2"):
         if song_name:
             title = song_name
 
-        print("audio2midi OK")
         xml = midi2xml(midi, title)
-        print("midi2xml OK")
         abc = xml2abc(xml)
-        print("xml2abc OK")
         mxl = xml2mxl(xml)
-        print("xml2mxl OK")
         pdf, jpg = xml2jpg(xml)
-        print("xml2jpg OK")
         return audio_path, midi, pdf, xml, mxl, abc, jpg
 
     except Exception as e:

requirements.txt

@@ -1,6 +1,7 @@
-torch
 pymupdf
 music21
 piano_transcription_inference
-modelscope[framework]==1.18
-librosa==0.9.2
+modelscope==1.15
+librosa==0.9.2
+torch==1.12.1
+numpy==1.26.4