720 sentences - plots shadow

visualize_tts_plesantness.py

@@ -10,12 +10,14 @@
 # mimic3_770.wav
 # mimic3_speedup_770.wav
 FULL_WAV  = [
-    'english_z.wav',
-    'english_4x_z.wav',
-    'human_z.wav',
-    'foreign_z.wav',
-    'foreign_4x_z.wav',
+    'english_hfullh.wav',
+    'english_4x_hfullh.wav',
+    'human_hfullh.wav',
+    'foreign_hfullh.wav',
+    'foreign_4x_hfullh.wav',
                     ]
+WIN = 40
+HOP = 10
 import pandas as pd
 import os
 
@@ -231,8 +233,8 @@ for long_audio in FULL_WAV:
             process_func=process_function,
             # process_func_args={'outputs': 'logits_scene'},
             process_func_applies_sliding_window=False,
-            win_dur=4.0,
-            hop_dur=40.0,
+            win_dur=WIN,
+            hop_dur=HOP,
             sampling_rate=16000,
             resample=True,
             verbose=True,
@@ -284,36 +286,40 @@ for lang in ['english',
              'foreign']:
             
 
-            fig, ax = plt.subplots(nrows=8, ncols=2, figsize=(21, 24),
+            fig, ax = plt.subplots(nrows=8, ncols=2, figsize=(24,20.7),
                                    gridspec_kw={'hspace': 0, 'wspace': .04})
 
 
             
 
-            time_stamp = preds['human_z.wav'].index.to_numpy()
+            time_stamp = preds['human_hfullh.wav'].index.to_numpy()
             for j, dim in enumerate(['arousal', 
                                     'dominance', 
                                     'valence']):
 
                 # MIMIC3                      
 
-                ax[j, 0].plot(time_stamp, preds[f'{lang}_z.wav'][dim], 
+                ax[j, 0].plot(time_stamp, preds[f'{lang}_hfullh.wav'][dim], 
                             color=(0,104/255,139/255), 
                             label='mean_1', 
                             linewidth=2)
                 ax[j, 0].fill_between(time_stamp,
 
-                                preds[f'{lang}_z.wav'][dim],
-                                preds['human_z.wav'][dim],
+                                0*preds[f'{lang}_hfullh.wav'][dim],
+                                preds['human_hfullh.wav'][dim],
 
                                 color=(.2,.2,.2), 
                                 alpha=0.244)
-                if j == 0:                    
-                    ax[j, 0].legend([f'StyleTTS2 using {lang}',
-                                     f'StyleTTS2 uising LibriSpeech'], 
-                                    prop={'size': 10},
+                if j == 0:
+                    if lang == 'english':
+                        desc = 'English'
+                    else:
+                        desc = 'Non-English'
+                    ax[j, 0].legend([f'StyleTTS2 using Mimic-3 {desc}',
+                                     f'StyleTTS2 uising EmoDB'], 
+                                    prop={'size': 14},
                                     )
-                ax[j, 0].set_ylabel(dim.lower(), color=(.4, .4, .4), fontsize=14)
+                ax[j, 0].set_ylabel(dim.lower(), color=(.4, .4, .4), fontsize=17)
                 
                 # TICK
                 ax[j, 0].set_ylim([1e-7, .9999])
@@ -326,26 +332,30 @@ for lang in ['english',
             # MIMIC3   4x speed
 
 
-                ax[j, 1].plot(time_stamp, preds[f'{lang}_4x_z.wav'][dim], 
+                ax[j, 1].plot(time_stamp, preds[f'{lang}_4x_hfullh.wav'][dim], 
                             color=(0,104/255,139/255), 
                             label='mean_1', 
                             linewidth=2)
                 ax[j, 1].fill_between(time_stamp,
 
-                                preds[f'{lang}_4x_z.wav'][dim],
-                                preds['human_z.wav'][dim],
+                                0 * preds[f'{lang}_4x_hfullh.wav'][dim],
+                                preds['human_hfullh.wav'][dim],
 
-                                color=(.2,.2,.2), 
+                                color=(.2,.2,.2),
                                 alpha=0.244)
-                if j == 0:                    
-                    ax[j, 1].legend([f'StyleTTS2 using {lang} 4x speed',
-                                    f'StyleTTS2 using LibriSpeech'], 
-                                    prop={'size': 10}, 
+                if j == 0:
+                    if lang == 'english':
+                        desc = 'English'
+                    else:
+                        desc = 'Non-English'                 
+                    ax[j, 1].legend([f'StyleTTS2 using Mimic-3 {desc} 4x speed',
+                                    f'StyleTTS2 using EmoDB'], 
+                                    prop={'size': 14}, 
                                     #  loc='lower right'
                                     )
 
 
-                ax[j, 1].set_xlabel('767 Harvard Sentences (seconds)')
+                ax[j, 1].set_xlabel('720 Harvard Sentences')
 
 
 
@@ -366,7 +376,7 @@ for lang in ['english',
 
 
 
-            time_stamp = preds['human_z.wav'].index.to_numpy()
+            time_stamp = preds['human_hfullh.wav'].index.to_numpy()
             for j, dim in enumerate(['Angry', 
                                     'Sad',
                                     'Happy',
@@ -380,14 +390,14 @@ for lang in ['english',
 
                 # MIMIC3                      
 
-                ax[j, 0].plot(time_stamp, preds[f'{lang}_z.wav'][dim], 
+                ax[j, 0].plot(time_stamp, preds[f'{lang}_hfullh.wav'][dim], 
                             color=(0,104/255,139/255), 
                             label='mean_1', 
                             linewidth=2)
                 ax[j, 0].fill_between(time_stamp,
 
-                                preds[f'{lang}_z.wav'][dim],
-                                preds['human_z.wav'][dim],
+                                0*preds[f'{lang}_hfullh.wav'][dim],
+                                preds['human_hfullh.wav'][dim],
 
                                 color=(.2,.2,.2), 
                                 alpha=0.244)
@@ -398,26 +408,26 @@ for lang in ['english',
                 # )
 
 
-                ax[j, 0].set_ylabel(dim.lower(), color=(.4, .4, .4), fontsize=14)
+                ax[j, 0].set_ylabel(dim.lower(), color=(.4, .4, .4), fontsize=17)
 
                 # TICKS
                 ax[j, 0].set_ylim([1e-7, .9999])
                 ax[j, 0].set_xlim([time_stamp[0], time_stamp[-1]])
                 ax[j, 0].set_xticklabels(['' for _ in ax[j, 0].get_xticklabels()])
-                ax[j, 0].set_xlabel('767 Harvard Sentences (seconds)', fontsize=16, color=(.4,.4,.4))
+                ax[j, 0].set_xlabel('720 Harvard Sentences', fontsize=17, color=(.2,.2,.2))
 
 
             # MIMIC3   4x speed
 
 
-                ax[j, 1].plot(time_stamp, preds[f'{lang}_4x_z.wav'][dim],
+                ax[j, 1].plot(time_stamp, preds[f'{lang}_4x_hfullh.wav'][dim],
                             color=(0,104/255,139/255), 
                             label='mean_1', 
                             linewidth=2)
                 ax[j, 1].fill_between(time_stamp,
 
-                                preds[f'{lang}_4x_z.wav'][dim],
-                                preds['human_z.wav'][dim],
+                                0*preds[f'{lang}_4x_hfullh.wav'][dim],
+                                preds['human_hfullh.wav'][dim],
 
                                 color=(.2,.2,.2), 
                                 alpha=0.244)
@@ -426,8 +436,8 @@ for lang in ['english',
                 #                  prop={'size': 10},
                 #                 #  loc='upper left'
                 # )
-                ax[j, 1].set_xlabel('767 Harvard Sentences (seconds)', fontsize=16, color=(.4,.4,.4))
-                ax[j, 1].set_ylim([1e-7, .999])
+                ax[j, 1].set_xlabel('720 Harvard Sentences', fontsize=17, color=(.2,.2,.2))
+                ax[j, 1].set_ylim([1e-7, .9999])
                 # ax[j, 1].set_yticklabels(['' for _ in ax[j, 1].get_yticklabels()])
                 ax[j, 1].set_xticklabels(['' for _ in ax[j, 1].get_xticklabels()])
                 ax[j, 1].set_xlim([time_stamp[0], time_stamp[-1]])
@@ -442,6 +452,6 @@ for lang in ['english',
 
 
 
-            plt.savefig(f'fig_{lang}_z.pdf', bbox_inches='tight')
+            plt.savefig(f'fig_{lang}_{WIN=}_{HOP=}_fin0.pdf', bbox_inches='tight')
             plt.close()