Update app.py

app.py

@@ -59,7 +59,7 @@ for m in tqdm.tqdm(melody_chords_f):
         for tv in range(-3, 3):
         
             section = str_strip(m[2])
-            section_tok = sections_dict.index(part)
+            section_tok = sections_dict.index(section)
             
             score = [t+tv if 256 < t < 512 else t for t in m[5]]
 
@@ -149,32 +149,32 @@ def Generate_POP_Section(input_parsons_code,
 
     print('Generating melody...')
 
-    song = []
-
-    song.append(389)
-    song.extend(pcode)
-    song.append(390)
-    
-    song.extend([388, 0, 10+128, 66+256])
-    
-    for i in tqdm.tqdm(range(1, len(td_str[:64]))):
+    # seq = [916] + [part_tok+512, mode_time(score)+532, mode_dur(score)+660, mode_pitch(score)+788]
     
-        song.append(pcode[i])
+    part = 'chorus'
+    mode_time = 10
+    mode_dur = 10
+    mode_pitch = 66
     
-        x = torch.tensor(song, dtype=torch.long, device='cpu')
+    #seq = [916] + [parts_dict.index(part)+512, mode_time+532, mode_dur+660, mode_pitch+788]
     
-        with ctx:
-            out = model.generate(x,
-                                 3,
-                                 filter_logits_fn=top_k,
-                                 filter_kwargs={'k': 1},
-                                 temperature=1.0,
-                                 return_prime=False,
-                                 verbose=False)
+    #seq = random.choice(train_data)[:128]
     
-        y = out.tolist()[0]
+    x = torch.LongTensor(seq).cuda()
     
-        song.extend(y)
+    with ctx:
+        out = model.generate(x,
+                             512,
+                             temperature=0.9,
+                             #filter_logits_fn=top_p,
+                             #filter_kwargs={'thres': 0.96},
+                             eos_token=917,
+                             return_prime=True,
+                             verbose=True)
+
+    y = out.tolist()[0]
+
+    song.extend(y)
 
     print('Done!')
     print('=' * 70)
@@ -287,49 +287,52 @@ if __name__ == "__main__":
             "Check out [popular-hook](https://huggingface.co/datasets/NEXTLab-ZJU/popular-hook) on Hugging Face!\n\n"
         )
         
-        gr.Markdown("## Select generation options:")
-        
-        input_parsons_code = gr.Textbox(label="Parsons code",
-                                        info="Make sure your Parsons code starts with *",
-                                        lines=1,
-                                        value="*"
-                                        )
+        gr.Markdown("## Select POP composition section to generate:")
+
+        input_comp_section = gr.Dropdown(sections_dict + ['random'], label="Composition section", value='intro')
         
         gr.Markdown("## Select generation options:")
-                
-        input_first_note_duration = gr.Slider(1, 127, value=15, step=1, label="First note duration value")
-        iinput_first_note_MIDI_pitch = gr.Slider(1, 127, value=60, step=1, label="First note MIDI pitch")
-         
-        run_btn = gr.Button("Generate melody", variant="primary")
+
+        input_mode_time = gr.Slider(0, 127, value=10, step=1, label="Composition mode time")
+        input_mode_dur = gr.Slider(0, 127, value=15, step=1, label="Composition mode dur")
+        input_mode_ptc = gr.Slider(0, 127, value=60, step=1, label="Composition mode pitch")
+        
+        run_btn = gr.Button("Generate", variant="primary")
         
         gr.Markdown("## Output results")
 
         output_midi_title = gr.Textbox(label="Output MIDI title")
+        output_midi_summary = gr.Textbox(label="Output MIDI summary")
         output_audio = gr.Audio(label="Output MIDI audio", format="mp3", elem_id="midi_audio")
         output_plot = gr.Plot(label="Output MIDI score plot")
         output_midi = gr.File(label="Output MIDI file", file_types=[".mid"])
 
-        run_event = run_btn.click(Generate_Melody, [input_parsons_code, 
-                                                   input_first_note_duration, 
-                                                   iinput_first_note_MIDI_pitch
-                                                  ],
-                                                [output_midi_title, output_midi, output_audio, output_plot])
-
-        gr.Examples(
-            [["*UUUUUUUDDDDDDDUUUUUUUDDDDDDDUUUUUUUDDDDDDDUUUUUUUDDDDDDDUUUUUUU", 15, 60],
-             ["*UDDDUDDDUDRURUDUUDRDDUDDRUDUDURUDRUDUDDDUDDDRDUURUDUUDDDUDRRUUD", 15, 60],
-             ["*DUDDDUUDDUUDDUDUDDDUUUUUDDDDUDDDUUDDUUDDUUDUDDUDDDUUDDUUDDUDUDD", 15, 60],
-             ["*DUUDDRDDUURUDUDDDUDDDDDURDDUDRDURUURUURDDDUURDUURUDUUDURDUDUDRD", 15, 60],
-             ["*UUUDDUUUDDDDDUDDUUDDDDUUDDUDDDDDUUUDDDDDUDDUUUDDDURDUDUUUDDUUUD", 15, 60],
-             ["*UDUUDRUDDUDRURUURUUUUUDUDDUDDUDDUDRUDDUDRUDDDUDUUDRUDDUDRURUURU", 15, 60],
-            ],
-            [input_parsons_code, 
-            input_first_note_duration, 
-            iinput_first_note_MIDI_pitch
-            ],
-            [output_midi_title, output_midi, output_audio, output_plot],
-            Generate_Melody,
-            cache_examples=True,
-        )
+        run_event = run_btn.click(Generate_POP_Section, [input_comp_section, 
+                                                        input_mode_time, 
+                                                        input_mode_dur,
+                                                        input_mode_ptc
+                                                        ],
+                                                        [output_midi_title,
+                                                         output_midi_summary,
+                                                         output_midi, 
+                                                         output_audio, 
+                                                         output_plot]
+                                                        )
+
+        gr.Examples([["intro", 10, 15, 60],
+                    ],
+                    [input_comp_section, 
+                    input_mode_time, 
+                    input_mode_dur,
+                    input_mode_ptc
+                    ],
+                    [output_midi_title,
+                    output_midi_summary,
+                    output_midi, 
+                    output_audio, 
+                    output_plot],
+                    Generate_POP_Section,
+                    cache_examples=True,
+                    )
         
         app.queue().launch()