Update app.py

app.py

@@ -13,19 +13,10 @@ import commons
 import utils
 from data_utils import TextAudioLoader, TextAudioCollate, TextAudioSpeakerLoader, TextAudioSpeakerCollate
 from models import SynthesizerTrn
-#from text.symbols import symbols
-#from text.symbols_gra import symbols
+from text.symbols import symbols
 from text import text_to_sequence
 from scipy.io.wavfile import write
 
-########################################
-_pad        = '_'
-_punctuation = ';:,.!?¡¿—-<>*()…"«»“”~ '
-_letters = 'ABCÇDEFGHIJKLMNOPQRSTUVWXYZÂÊÎÔÛâêîôûéÉèåÅÈàÀùÙÌìëöabcçdefghijklmnopqrstuvwxyz'
-_letters_ipa = "ɑɐɒæɓʙβɔɕçɗɖðʤəɘɚɛɜɝɞɟʄɡɠɢʛɦɧħɥʜɨɪʝɭɬɫɮʟɱɯɰŋɳɲɴøɵɸθœɶʘɹɺɾɻʀʁɽʂʃʈʧʉʊʋⱱʌɣɤʍχʎʏʑʐʒʔʡʕʢǀǁǂǃˈˌːˑʼʴʰʱʲʷˠˤ˞↓↑→↗↘'̩'ᵻ"
-symbols = [_pad] + list(_punctuation) + list(_letters) + list(_letters_ipa)
-SPACE_ID = symbols.index(" ")
-
 def get_text(text, hps):
     text_norm = text_to_sequence(text, hps.data.text_cleaners)
     if hps.data.add_blank:
@@ -45,18 +36,15 @@ def load_model(model_path, hps):
     return net_g
 
 #hps = utils.get_hparams_from_file("configs/vctk_base.json")
-hps = utils.get_hparams_from_file("wa_graphemes/config.json")
+hps = utils.get_hparams_from_file("configs/vctk_base.json")
 
 # Define a dictionary to store the model paths for each tab
 model_paths = {
-    "Phonemes_finetuned": "fr_wa_finetuned_pho/G_125000.pth",
-    "Graphemes_finetuned": "fr_wa_finetuned/G_198000.pth",
-    "Phonemes": "path_to_phonemes_model.pth",
-    "Graphemes": "wa_graphemes/G_258000.pth"
+    "Phonemes_finetuned": "fr_wa_finetuned_pho/G_125000.pth"
 }
 
 # Load the initial model
-net_g = load_model(model_paths["Graphemes"], hps)
+net_g = load_model(model_paths["Phonemes_finetuned"], hps)
 
 def tts(text, speaker_id, tab_name):
     global net_g
@@ -74,7 +62,7 @@ def tts(text, speaker_id, tab_name):
 def create_tab(tab_name):
     with gr.TabItem(tab_name):
         gr.Markdown(f"### {tab_name} TTS Model")
-        tts_input1 = gr.TextArea(label="Text in Walloon (Depending on the model the input should be on phonemes or characters)", value="")
+        tts_input1 = gr.TextArea(label="Text in Walloon on IPA phonemes", value="")
         tts_input2 = gr.Dropdown(label="Speaker", choices=["Male", "Female"], type="index", value="Male")
         tts_submit = gr.Button("Generate", variant="primary")
         tts_output1 = gr.Textbox(label="Message")
@@ -88,16 +76,13 @@ with app:
         # First Text to Speech (TTS) for Walloon
         Based on VITS (https://github.com/jaywalnut310/vits).
 
-        Select the desired model and write the text in phonemes or graphemes depending on the model.
+        Write the text in phonemes or graphemes depending on the model.
 
         For faster inference speed it is recommended to use short sentences.
         """
     )
     with gr.Tabs():
         create_tab("Phonemes_finetuned")
-        create_tab("Graphemes_finetuned")
-        create_tab("Phonemes")
-        create_tab("Graphemes")
 
     gr.Markdown(
         """