Update app.py

app.py

@@ -1,32 +1,25 @@
 INTROTXT = """# StyleTTS 2
+kudos to mrfakename for the base gradio code I'm borrowing here.
 
-[Paper](https://arxiv.org/abs/2306.07691) - [Samples](https://styletts2.github.io/) - [Code](https://github.com/yl4579/StyleTTS2) - [Discord](https://discord.gg/ha8sxdG2K4)
-
-A free demo of StyleTTS 2. **I am not affiliated with the StyleTTS 2 Authors.**
-
-**Before using this demo, you agree to inform the listeners that the speech samples are synthesized by the pre-trained models, unless you have the permission to use the voice you synthesize. That is, you agree to only use voices whose speakers grant the permission to have their voice cloned, either directly or by license before making synthesized voices public, or you have to publicly announce that these voices are synthesized if you do not have the permission to use these voices.**
-
-Is there a long queue on this space? Duplicate it and add a more powerful GPU to skip the wait! **Note: Thank you to Hugging Face for their generous GPU grant program!**
-
-**NOTE: StyleTTS 2 does better on longer texts.** For example, making it say "hi" will produce a lower-quality result than making it say a longer phrase.
-
-**NOTE: StyleTTS 2 is _currently_ English-only. Join the Discord for updates on multilingual training.**
+日本語用
+The Text-guided inference may or may not work. you can only do inference max 512 tokens.
+**
 """
 import gradio as gr
 import styletts2importable
-import ljspeechimportable
+import Text-guided Inferenceimportable
 import torch
 import os
 from txtsplit import txtsplit
 import numpy as np
 import pickle
-theme = gr.themes.Base(
+theme = gr.themes.Base(theme="NoCrypt/miku",
     font=[gr.themes.GoogleFont('Libre Franklin'), gr.themes.GoogleFont('Public Sans'), 'system-ui', 'sans-serif'],
 )
-voicelist = ['f-us-1', 'f-us-2', 'f-us-3', 'f-us-4', 'm-us-1', 'm-us-2', 'm-us-3', 'm-us-4']
+voicelist = ['1','2','3']
 voices = {}
-import phonemizer
-global_phonemizer = phonemizer.backend.EspeakBackend(language='en-us', preserve_punctuation=True,  with_stress=True)
+# import phonemizer
+# global_phonemizer = phonemizer.backend.EspeakBackend(language='en-us', preserve_punctuation=True,  with_stress=True)
 # todo: cache computed style, load using pickle
 # if os.path.exists('voices.pkl'):
     # with open('voices.pkl', 'rb') as f:
@@ -108,7 +101,7 @@ def ljsynthesize(text, steps, progress=gr.Progress()):
     # if len(text) > 400:
     #     raise gr.Error("Text must be under 400 characters")
     noise = torch.randn(1,1,256).to('cuda' if torch.cuda.is_available() else 'cpu')
-    # return (24000, ljspeechimportable.inference(text, noise, diffusion_steps=7, embedding_scale=1))
+    # return (24000, Text-guided Inferenceimportable.inference(text, noise, diffusion_steps=7, embedding_scale=1))
     if text.strip() == "":
         raise gr.Error("You must enter some text")
     if len(text) > 150000:
@@ -119,7 +112,7 @@ def ljsynthesize(text, steps, progress=gr.Progress()):
     texts = txtsplit(text)
     audios = []
     for t in progress.tqdm(texts):
-        audios.append(ljspeechimportable.inference(t, noise, diffusion_steps=steps, embedding_scale=1))
+        audios.append(Text-guided Inferenceimportable.inference(t, noise, diffusion_steps=steps, embedding_scale=1))
     return (24000, np.concatenate(audios))
 
 
@@ -128,7 +121,7 @@ with gr.Blocks() as vctk:
         with gr.Column(scale=1):
             inp = gr.Textbox(label="Text", info="What would you like StyleTTS 2 to read? It works better on full sentences.", interactive=True)
             voice = gr.Dropdown(voicelist, label="Voice", info="Select a default voice.", value='m-us-2', interactive=True)
-            multispeakersteps = gr.Slider(minimum=3, maximum=15, value=3, step=1, label="Diffusion Steps", info="Theoretically, higher should be better quality but slower, but we cannot notice a difference. Try with lower steps first - it is faster", interactive=True)
+            multispeakersteps = gr.Slider(minimum=3, maximum=15, value=3, step=1, label="Diffusion Steps", info="Higher gives you more diverse results but not necessarily higher quality - これを増えたらもっとエモーショナルな結果になりますが、クオリティーのいい結果になるとは限らない。", interactive=True)
             # use_gruut = gr.Checkbox(label="Use alternate phonemizer (Gruut) - Experimental")
         with gr.Column(scale=1):
             btn = gr.Button("Synthesize", variant="primary")
@@ -170,8 +163,8 @@ with gr.Blocks() as lj:
 with gr.Blocks(title="StyleTTS 2", css="footer{display:none !important}", theme=theme) as demo:
     gr.Markdown(INTROTXT)
     gr.DuplicateButton("Duplicate Space")
-    # gr.TabbedInterface([vctk, clone, lj, longText], ['Multi-Voice', 'Voice Cloning', 'LJSpeech', 'Long Text [Beta]'])
-    gr.TabbedInterface([vctk, clone, lj], ['Multi-Voice', 'Voice Cloning', 'LJSpeech', 'Long Text [Beta]'])
+    # gr.TabbedInterface([vctk, clone, lj, longText], ['Multi-Voice', 'Voice Cloning', 'Text-guided Inference', 'Long Text [Beta]'])
+    gr.TabbedInterface([vctk, clone, lj], ['Multi-Voice', 'Voice Cloning (don't use this,このオプションはこのモデルに利きません。', 'Text-guided Inference', 'Long Text [Beta]'])
     gr.Markdown("""
 Demo by [mrfakename](https://twitter.com/realmrfakename). I am not affiliated with the StyleTTS 2 authors.