Upload 5 files

app.py

@@ -0,0 +1,52 @@
+import gradio as gr
+import numpy
+import librosa
+from pydub import *
+import fastbook
+from fastbook import *
+import pathlib
+temp = pathlib.PosixPath
+pathlib.PosixPath = pathlib.WindowsPath
+model = load_learner("thedrum.pkl")
+
+def play(x,y,ti,beat,audio):
+
+  sts=0
+  b=0
+  all=[]
+  countfname=1
+  for i in range(int(ti*beat*2)):  
+    sound = AudioSegment.from_mp3(audio)
+    StrtSec = sts
+    EndSec = beat*(i+1)/2
+    StrtTime = StrtSec*1000
+    EndTime = EndSec*1000
+    extract = sound[StrtTime:EndTime]
+    extract.export("Half.wav", format="wav")
+
+      
+
+    x,y = librosa.load('Half.wav')
+    plt.figure(figsize=(12,4))
+    a = librosa.feature.melspectrogram(y=x,sr=y,n_mels=550)
+    b = librosa.power_to_db(a,ref=np.max)
+    librosa.display.specshow(b,sr=y, x_axis='time', y_axis='mel')
+    plt.savefig(f'{countfname}')
+    wit = model.predict(f'{countfname}.png')
+    all.append(wit[0])
+    countfname+=1
+
+    sts=EndSec
+  return all
+
+def transcribe(audio):
+    x,y = librosa.load(audio)
+    ti = librosa.get_duration(y=x,sr=y)
+    beat = 1
+    text = play(x,y,ti,beat,audio)
+    return text
+
+gr.Interface(
+    fn=transcribe, 
+    inputs=gr.Audio(type="filepath"), 
+    outputs="text").launch()

thedrum.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:957dedb8f9fd2a9648b5564f4cf168d83a121624eede0091b67557c0465b1de9
+size 87805227