create new advanced control GUI

InferenceInterfaces/ToucanTTSInterface.py

@@ -144,7 +144,8 @@ def forward(self,
                 input_is_phones=False,
                 return_plot_as_filepath=False,
                 loudness_in_db=-29.0,
-                prosody_creativity=0.1):
+                prosody_creativity=0.1,
+                return_everything=False):
         """
         duration_scaling_factor: reasonable values are 0.8 < scale < 1.2.
                                      1.0 means no scaling happens, higher values increase durations for the whole
@@ -233,6 +234,8 @@ def forward(self,
                 plt.savefig("tmp.png")
                 plt.close()
                 return wave, sr, "tmp.png"
+        if return_everything:
+            return wave, mel, durations, pitch
         return wave, sr
 
     def read_to_file(self,

Modules/ToucanTTS/InferenceToucanTTS.py

@@ -242,7 +242,7 @@ def _forward(self,
                                                                              mask=text_masks.float(),
                                                                              n_timesteps=20,
                                                                              temperature=prosody_creativity,
-                                                                             c=utterance_embedding)), min=0.0).long().squeeze(1) if gold_durations is None else gold_durations
+                                                                             c=utterance_embedding)), min=0.0).long().squeeze(1) if gold_durations is None else gold_durations.squeeze(1)
 
         # modifying the predictions with control parameters
         for phoneme_index, phoneme_vector in enumerate(text_tensors.squeeze(0)):