tee342 commited on
Commit
c1c8f00
ยท
verified ยท
1 Parent(s): 65bee5e

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +148 -20
app.py CHANGED
@@ -23,6 +23,9 @@ import base64
23
  import pickle
24
  import json
25
  import soundfile as sf
 
 
 
26
 
27
  # Suppress warnings
28
  warnings.filterwarnings("ignore")
@@ -203,7 +206,7 @@ def stem_split(audio_path):
203
  stem_paths = []
204
 
205
  for i, name in enumerate(['drums', 'bass', 'other', 'vocals']):
206
- path = os.path.join(output_dir, f"{name}.wav")
207
  save_track(path, sources[i].cpu(), model.samplerate)
208
  stem_paths.append(gr.File(value=path))
209
 
@@ -254,7 +257,7 @@ def process_audio(audio_file, selected_effects, isolate_vocals, preset_name, exp
254
 
255
  waveform_image = show_waveform(output_path)
256
  genre = detect_genre(output_path)
257
- session_log = generate_session_log(audio_file, effects_to_apply, isolate_vocals, export_format, genre)
258
 
259
  status = "๐ŸŽ‰ Done!"
260
  return output_path, waveform_image, session_log, genre, status, history
@@ -526,12 +529,81 @@ def formant_correct(audio, shift=1.0):
526
  corrected = librosa.effects.pitch_shift(samples, sr=sr, n_steps=shift)
527
  return array_to_audiosegment(corrected.astype(np.int16), sr, channels=audio.channels)
528
 
529
- # === Effect Chain Sharing Functions ===
530
- def share_effect_chain(selected_effects):
531
- return json.dumps(selected_effects)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
532
 
533
- def load_effect_chain(chain_json):
534
- return json.loads(chain_json)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
535
 
536
  # === Main UI โ€“ With Studio Pulse Branding ===
537
  with gr.Blocks(css="""
@@ -721,7 +793,7 @@ with gr.Blocks(css="""
721
  output_other = gr.Audio(label="Processed Other", type="filepath")
722
  btn_other.click(fn=apply_effects_to_stem, inputs=[other_path, other_effects], outputs=output_other)
723
 
724
- # --- AI Remastering Tab โ€“ Now Fixed & Working ===
725
  with gr.Tab("๐Ÿ”ฎ AI Remastering"):
726
  gr.Interface(
727
  fn=ai_remaster,
@@ -742,8 +814,7 @@ with gr.Blocks(css="""
742
  ],
743
  outputs=gr.Audio(label="Warm Output", type="filepath"),
744
  title="Add Analog-Style Warmth",
745
- description="Enhance clarity and presence using saturation styles like Tube or Tape.",
746
- allow_flagging="never"
747
  )
748
 
749
  # --- Vocal Doubler / Harmonizer โ€“ Added ===
@@ -796,7 +867,6 @@ with gr.Blocks(css="""
796
  y_np = y.numpy().flatten()
797
  stft = librosa.stft(y_np)
798
  db = librosa.amplitude_to_db(abs(stft))
799
-
800
  plt.figure(figsize=(10, 4))
801
  img = librosa.display.specshow(db, sr=sr, x_axis="time", y_axis="hz", cmap="magma")
802
  plt.colorbar(img, format="%+2.0f dB")
@@ -812,9 +882,7 @@ with gr.Blocks(css="""
812
  gr.Interface(
813
  fn=visualize_spectrum,
814
  inputs=gr.Audio(label="Upload Track", type="filepath"),
815
- outputs=gr.Image(label="Spectrum Analysis"),
816
- title="Real-Time Spectrum Analyzer",
817
- description="See the frequency breakdown of your audio"
818
  )
819
 
820
  # --- Loudness Graph Tab โ€“ EBU R128 Matching ===
@@ -970,7 +1038,7 @@ with gr.Blocks(css="""
970
  allow_flagging="never"
971
  )
972
 
973
- # --- Loop Section ===
974
  with gr.Tab("๐Ÿ” Loop Playback"):
975
  gr.Interface(
976
  fn=loop_section,
@@ -988,7 +1056,7 @@ with gr.Blocks(css="""
988
  # --- Share Effect Chain Tab โ€“ Now Defined! ===
989
  with gr.Tab("๐Ÿ”— Share Effect Chain"):
990
  gr.Interface(
991
- fn=share_effect_chain,
992
  inputs=gr.CheckboxGroup(choices=preset_choices["Default"]),
993
  outputs=gr.Textbox(label="Share Code", lines=2),
994
  title="Copy/Paste Effect Chain",
@@ -997,7 +1065,7 @@ with gr.Blocks(css="""
997
 
998
  with gr.Tab("๐Ÿ“ฅ Load Shared Chain"):
999
  gr.Interface(
1000
- fn=load_effect_chain,
1001
  inputs=gr.Textbox(label="Paste Shared Code", lines=2),
1002
  outputs=gr.CheckboxGroup(choices=preset_choices["Default"], label="Loaded Effects"),
1003
  title="Restore From Shared Chain",
@@ -1010,14 +1078,14 @@ with gr.Blocks(css="""
1010
  ### Keyboard Controls
1011
  - `Ctrl + Z`: Undo last effect
1012
  - `Ctrl + Y`: Redo
1013
- - `Spacebar`: Play/Stop
1014
  - `Ctrl + S`: Save current session
1015
  - `Ctrl + O`: Open session
1016
  - `Ctrl + C`: Copy effect chain
1017
  - `Ctrl + V`: Paste effect chain
1018
  - `Ctrl + Shift + L`: Loop selection
1019
- - `Ctrl + Shift + P`: Export PDF report
1020
- - `Ctrl + Shift + M`: Export full mix ZIP
1021
  """)
1022
 
1023
  # --- Vocal Formant Correction โ€“ Now Defined! ===
@@ -1033,4 +1101,64 @@ with gr.Blocks(css="""
1033
  description="Make pitch-shifted vocals sound more human"
1034
  )
1035
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1036
  demo.launch()
 
23
  import pickle
24
  import json
25
  import soundfile as sf
26
+ import wave
27
+ import struct
28
+ from fpdf import FPDF
29
 
30
  # Suppress warnings
31
  warnings.filterwarnings("ignore")
 
206
  stem_paths = []
207
 
208
  for i, name in enumerate(['drums', 'bass', 'other', 'vocals']):
209
+ path = os.path.join(outputfile_dir, f"{name}.wav")
210
  save_track(path, sources[i].cpu(), model.samplerate)
211
  stem_paths.append(gr.File(value=path))
212
 
 
257
 
258
  waveform_image = show_waveform(output_path)
259
  genre = detect_genre(output_path)
260
+ session_log = generate_session_log(audio_file, selected_effects, isolate_vocals, export_format, genre)
261
 
262
  status = "๐ŸŽ‰ Done!"
263
  return output_path, waveform_image, session_log, genre, status, history
 
529
  corrected = librosa.effects.pitch_shift(samples, sr=sr, n_steps=shift)
530
  return array_to_audiosegment(corrected.astype(np.int16), sr, channels=audio.channels)
531
 
532
+ # === DAW Template Generator ===
533
+ def generate_ableton_template(stems):
534
+ # Simulate template metadata
535
+ template = {
536
+ "format": "Ableton Live",
537
+ "stems": [os.path.basename(s) for s in stems],
538
+ "effects": ["Reverb", "EQ", "Compression"],
539
+ "tempo": 128,
540
+ "title": "Studio Pulse Project"
541
+ }
542
+ out_path = os.path.join(tempfile.gettempdir(), "ableton_template.json")
543
+ with open(out_path, "w") as f:
544
+ json.dump(template, f, indent=2)
545
+ return out_path
546
+
547
+ # === Multi-stem ZIP Export ===
548
+ def export_full_mix(stems, final_mix):
549
+ zip_path = os.path.join(tempfile.gettempdir(), "full_export.zip")
550
+ with zipfile.ZipFile(zip_path, "w") as zipf:
551
+ for i, stem in enumerate(stems):
552
+ zipf.write(stem, f"stem_{i}.wav")
553
+ zipf.write(final_mix, "final_mix.wav")
554
+ return zip_path
555
+
556
+ # === Export Session Report (PDF) ===
557
+ class PDFReport(FPDF):
558
+ def header(self):
559
+ self.set_font('Arial', 'B', 12)
560
+ self.cell(0, 10, 'Studio Pulse Session Report', ln=True, align='C')
561
+
562
+ def export_session_report(session_data):
563
+ pdf = PDFReport()
564
+ pdf.add_page()
565
+ pdf.set_font("Arial", size=10)
566
+ pdf.multi_cell(0, 10, txt=session_data)
567
+ out_path = os.path.join(tempfile.gettempdir(), "session_report.pdf")
568
+ pdf.output(out_path)
569
+ return out_path
570
 
571
+ # === AI Sound Design (Text-to-Sound) ===
572
+ def text_to_sound(prompt):
573
+ # Placeholder logic
574
+ tts = TTS(model="tts_models/en/vctk/vits")
575
+ out_path = os.path.join(tempfile.gettempdir(), "generated_sound.wav")
576
+ tts.tts_to_file(text=prompt, speaker="p225", file_path=out_path)
577
+ return out_path
578
+
579
+ # === Video Sync Mode ===
580
+ def extract_audio_from_video(video_path):
581
+ clip = VideoFileClip(video_path)
582
+ out_path = os.path.join(tempfile.gettempdir(), "video_audio.wav")
583
+ clip.audio.write_audiofile(out_path)
584
+ return out_path
585
+
586
+ def re_sync_audio(video_path, edited_audio_path):
587
+ original_clip = VideoFileClip(video_path)
588
+ edited_audio = AudioSegment.from_file(edited_audio_path)
589
+ temp_audio_path = os.path.join(tempfile.gettempdir(), "temp_edited.mp3")
590
+ edited_audio.export(temp_audio_path, format="mp3")
591
+ final_clip = original_clip.set_audio(temp_audio_path)
592
+ out_path = os.path.join(tempfile.gettempdir(), "final_video.mp4")
593
+ final_clip.write_videofile(out_path, codec="libx264", audio_codec="aac")
594
+ return out_path
595
+
596
+ # === Sidechain Compression / Ducking ===
597
+ def sidechain_ducking(background, voice, threshold=-16, ratio=4):
598
+ bg_samples, bg_sr = audiosegment_to_array(background)
599
+ voice_samples, v_sr = audiosegment_to_array(voice)
600
+ if bg_sr != v_sr:
601
+ voice_samples = librosa.resample(voice_samples, orig_sr=v_sr, target_sr=bg_sr)
602
+ bg_rms = np.sqrt(np.mean(bg_samples ** 2))
603
+ voice_rms = np.sqrt(np.mean(voice_samples ** 2))
604
+ if voice_rms > bg_rms * threshold / 100:
605
+ bg_samples = bg_samples * ratio
606
+ return array_to_audiosegment(bg_samples.astype(np.int16), bg_sr, channels=background.channels)
607
 
608
  # === Main UI โ€“ With Studio Pulse Branding ===
609
  with gr.Blocks(css="""
 
793
  output_other = gr.Audio(label="Processed Other", type="filepath")
794
  btn_other.click(fn=apply_effects_to_stem, inputs=[other_path, other_effects], outputs=output_other)
795
 
796
+ # --- AI Remastering Tab โ€“ New Feature (Fixed) ===
797
  with gr.Tab("๐Ÿ”ฎ AI Remastering"):
798
  gr.Interface(
799
  fn=ai_remaster,
 
814
  ],
815
  outputs=gr.Audio(label="Warm Output", type="filepath"),
816
  title="Add Analog-Style Warmth",
817
+ description="Enhance clarity and presence using saturation styles like Tube or Tape."
 
818
  )
819
 
820
  # --- Vocal Doubler / Harmonizer โ€“ Added ===
 
867
  y_np = y.numpy().flatten()
868
  stft = librosa.stft(y_np)
869
  db = librosa.amplitude_to_db(abs(stft))
 
870
  plt.figure(figsize=(10, 4))
871
  img = librosa.display.specshow(db, sr=sr, x_axis="time", y_axis="hz", cmap="magma")
872
  plt.colorbar(img, format="%+2.0f dB")
 
882
  gr.Interface(
883
  fn=visualize_spectrum,
884
  inputs=gr.Audio(label="Upload Track", type="filepath"),
885
+ outputs=gr.Image(label="Spectrum Analysis")
 
 
886
  )
887
 
888
  # --- Loudness Graph Tab โ€“ EBU R128 Matching ===
 
1038
  allow_flagging="never"
1039
  )
1040
 
1041
+ # --- Loop Playback ===
1042
  with gr.Tab("๐Ÿ” Loop Playback"):
1043
  gr.Interface(
1044
  fn=loop_section,
 
1056
  # --- Share Effect Chain Tab โ€“ Now Defined! ===
1057
  with gr.Tab("๐Ÿ”— Share Effect Chain"):
1058
  gr.Interface(
1059
+ fn=lambda x: json.dumps(x),
1060
  inputs=gr.CheckboxGroup(choices=preset_choices["Default"]),
1061
  outputs=gr.Textbox(label="Share Code", lines=2),
1062
  title="Copy/Paste Effect Chain",
 
1065
 
1066
  with gr.Tab("๐Ÿ“ฅ Load Shared Chain"):
1067
  gr.Interface(
1068
+ fn=json.loads,
1069
  inputs=gr.Textbox(label="Paste Shared Code", lines=2),
1070
  outputs=gr.CheckboxGroup(choices=preset_choices["Default"], label="Loaded Effects"),
1071
  title="Restore From Shared Chain",
 
1078
  ### Keyboard Controls
1079
  - `Ctrl + Z`: Undo last effect
1080
  - `Ctrl + Y`: Redo
1081
+ - `Spacebar`: Play/Stop playback
1082
  - `Ctrl + S`: Save current session
1083
  - `Ctrl + O`: Open session
1084
  - `Ctrl + C`: Copy effect chain
1085
  - `Ctrl + V`: Paste effect chain
1086
  - `Ctrl + Shift + L`: Loop selection
1087
+ - `Ctrl + Shift + P`: Export to PDF report
1088
+ - `Ctrl + Shift + M`: Export full mix ZIP
1089
  """)
1090
 
1091
  # --- Vocal Formant Correction โ€“ Now Defined! ===
 
1101
  description="Make pitch-shifted vocals sound more human"
1102
  )
1103
 
1104
+ # --- DAW Template Export ===
1105
+ with gr.Tab("๐ŸŽ› DAW Template Export"):
1106
+ gr.Interface(
1107
+ fn=generate_ableton_template,
1108
+ inputs=[gr.File(label="Upload Stems", file_count="multiple")],
1109
+ outputs=gr.File(label="DAW Template (.json/.als/.flp)")
1110
+ )
1111
+
1112
+ # --- Export Full Mix ZIP ===
1113
+ with gr.Tab("๐Ÿ“ Export Full Mix ZIP"):
1114
+ gr.Interface(
1115
+ fn=export_full_mix,
1116
+ inputs=[
1117
+ gr.File(label="Stems", file_count="multiple"),
1118
+ gr.File(label="Final Mix")
1119
+ ],
1120
+ outputs=gr.File(label="Full Mix Archive (.zip)"),
1121
+ title="Export Stems + Final Mix Together",
1122
+ description="Perfect for sharing with producers or archiving"
1123
+ )
1124
+
1125
+ # --- AI Sound Design โ€“ Text to Audio Generation ===
1126
+ with gr.Tab("๐Ÿง  AI Sound Design"):
1127
+ gr.Interface(
1128
+ fn=text_to_sound,
1129
+ inputs=gr.Textbox(label="Describe a Sound", lines=5),
1130
+ outputs=gr.Audio(label="Generated Sound", type="filepath"),
1131
+ title="Generate Sounds from Text",
1132
+ description="Describe a sound โ†’ AI generates it"
1133
+ )
1134
+
1135
+ # --- Session Report Export (PDF) ===
1136
+ with gr.Tab("๐Ÿ“„ Session Report"):
1137
+ gr.Interface(
1138
+ fn=export_session_report,
1139
+ inputs=gr.Textbox(label="Session Data (JSON)", lines=10),
1140
+ outputs=gr.File(label="Download PDF Report"),
1141
+ title="Export Detailed Session Report",
1142
+ description="For professional use or sharing with collaborators"
1143
+ )
1144
+
1145
+ # --- Video Sync Mode โ€“ Now Included ===
1146
+ with gr.Tab("๐Ÿ“ฝ Video Sync Mode"):
1147
+ with gr.Row():
1148
+ with gr.Column(min_width=300):
1149
+ input_video = gr.Video(label="Upload Video", type="filepath")
1150
+ extract_btn = gr.Button("Extract Audio")
1151
+ extracted_audio = gr.Audio(label="Extracted Audio", type="filepath")
1152
+ effect_checkbox_video = gr.CheckboxGroup(choices=preset_choices["Default"], label="Apply Effects")
1153
+ process_btn_video = gr.Button("Process Audio")
1154
+ with gr.Column(min_width=300):
1155
+ processed_audio = gr.Audio(label="Edited Audio", type="filepath")
1156
+ re_sync_btn = gr.Button("Sync Back to Video")
1157
+ output_video = gr.Video(label="Final Video", type="filepath")
1158
+
1159
+ extract_btn.click(fn=extract_audio_from_video, inputs=input_video, outputs=extracted_audio)
1160
+ process_btn_video.click(fn=process_audio, inputs=[extracted_audio, effect_checkbox_video, gr.Checkbox(value=False), gr.Dropdown(value="Default"), gr.Dropdown(value="MP3")],
1161
+ outputs=[processed_audio])
1162
+ re_sync_btn.click(fn=re_sync_audio, inputs=[input_video, processed_audio], outputs=output_video)
1163
+
1164
  demo.launch()