Spaces:
Build error
Build error
juancopi81
commited on
Commit
•
1f1f657
1
Parent(s):
c0fa1b2
Add piano roll
Browse files
app.py
CHANGED
@@ -9,7 +9,7 @@ from pytube import YouTube
|
|
9 |
from pydub import AudioSegment
|
10 |
|
11 |
from inferencemodel import InferenceModel
|
12 |
-
from utils import upload_audio
|
13 |
|
14 |
import nest_asyncio
|
15 |
nest_asyncio.apply()
|
@@ -24,13 +24,13 @@ current_model = "mt3"
|
|
24 |
def change_model(model):
|
25 |
global current_model
|
26 |
global inference_model
|
|
|
|
|
27 |
checkpoint_path = f"/home/user/app/checkpoints/{model}/"
|
28 |
if model == current_model:
|
29 |
return
|
30 |
inference_model = InferenceModel(checkpoint_path, model)
|
31 |
current_model = model
|
32 |
-
print("Inferece model", inference_model)
|
33 |
-
print("Current model", current_model)
|
34 |
|
35 |
# Credits https://huggingface.co/spaces/rajesh1729/youtube-video-transcription-with-whisper
|
36 |
def get_audio(url):
|
@@ -69,9 +69,9 @@ def inference(yt_audio_path):
|
|
69 |
synth = note_seq.midi_synth.fluidsynth
|
70 |
array_of_floats = synth(est_ns, sample_rate=SAMPLE_RATE, sf2_path=SF2_PATH)
|
71 |
int16_data = note_seq.audio_io.float_samples_to_int16(array_of_floats)
|
72 |
-
|
73 |
|
74 |
-
return "./transcribed.mid", (SAMPLE_RATE, int16_data)
|
75 |
|
76 |
title = "Transcribe music from YouTube videos using Transformers."
|
77 |
description = """
|
@@ -107,7 +107,7 @@ with demo:
|
|
107 |
img = gr.Image(label="Thumbnail")
|
108 |
with gr.Row():
|
109 |
yt_audio = gr.Audio()
|
110 |
-
yt_audio_path = gr.Textbox()
|
111 |
|
112 |
link.change(fn=populate_metadata, inputs=link, outputs=[img, title, yt_audio, yt_audio_path])
|
113 |
|
@@ -117,10 +117,12 @@ with demo:
|
|
117 |
with gr.Row():
|
118 |
midi_file = gr.File()
|
119 |
midi_audio = gr.Audio()
|
|
|
|
|
120 |
|
121 |
btn.click(inference,
|
122 |
inputs=yt_audio_path,
|
123 |
-
outputs=[midi_file, midi_audio])
|
124 |
|
125 |
gr.Markdown(article)
|
126 |
|
|
|
9 |
from pydub import AudioSegment
|
10 |
|
11 |
from inferencemodel import InferenceModel
|
12 |
+
from utils import upload_audio, create_image_from_note_sequence
|
13 |
|
14 |
import nest_asyncio
|
15 |
nest_asyncio.apply()
|
|
|
24 |
def change_model(model):
|
25 |
global current_model
|
26 |
global inference_model
|
27 |
+
print("Inferece model", inference_model)
|
28 |
+
print("Current model", current_model)
|
29 |
checkpoint_path = f"/home/user/app/checkpoints/{model}/"
|
30 |
if model == current_model:
|
31 |
return
|
32 |
inference_model = InferenceModel(checkpoint_path, model)
|
33 |
current_model = model
|
|
|
|
|
34 |
|
35 |
# Credits https://huggingface.co/spaces/rajesh1729/youtube-video-transcription-with-whisper
|
36 |
def get_audio(url):
|
|
|
69 |
synth = note_seq.midi_synth.fluidsynth
|
70 |
array_of_floats = synth(est_ns, sample_rate=SAMPLE_RATE, sf2_path=SF2_PATH)
|
71 |
int16_data = note_seq.audio_io.float_samples_to_int16(array_of_floats)
|
72 |
+
piano_roll = create_image_from_note_sequence(est_ns)
|
73 |
|
74 |
+
return "./transcribed.mid", (SAMPLE_RATE, int16_data), piano_roll
|
75 |
|
76 |
title = "Transcribe music from YouTube videos using Transformers."
|
77 |
description = """
|
|
|
107 |
img = gr.Image(label="Thumbnail")
|
108 |
with gr.Row():
|
109 |
yt_audio = gr.Audio()
|
110 |
+
yt_audio_path = gr.Textbox(visible=False)
|
111 |
|
112 |
link.change(fn=populate_metadata, inputs=link, outputs=[img, title, yt_audio, yt_audio_path])
|
113 |
|
|
|
117 |
with gr.Row():
|
118 |
midi_file = gr.File()
|
119 |
midi_audio = gr.Audio()
|
120 |
+
with gr.Row():
|
121 |
+
piano_roll = gr.Image()
|
122 |
|
123 |
btn.click(inference,
|
124 |
inputs=yt_audio_path,
|
125 |
+
outputs=[midi_file, midi_audio, piano_roll])
|
126 |
|
127 |
gr.Markdown(article)
|
128 |
|