commit from

app.py

@@ -43,16 +43,23 @@ from scipy.io.wavfile import write
 import gradio as gr
 import scipy.io.wavfile
 import numpy as np
-import torchtext
+import re
 
-def get_text(text, hps):
-    text_norm = cleaned_text_to_sequence(text)
-    if hps.data.add_blank:
-        text_norm = commons.intersperse(text_norm, 0)
-    text_norm = torch.LongTensor(text_norm)
-    return text_norm
+jp_match = re.compile(r'.*?[ぁ|あ|ぃ|い|ぅ|う|ぇ|え|ぉ|お|か|が|き|ぎ|く|ぐ|け|げ|こ|ご|さ|ざ|し|じ|す|ず|せ|ぜ|そ|ぞ|た|だ|ち|ぢ|っ|つ|づ|て|で|と|ど|な|に|ぬ|ね|の|は|ば|ぱ|ひ|び|ぴ|ふ|ぶ|ぷ|へ|べ|ぺ|ほ|ぼ|ぽ|ま|み|む|め|も|ゃ|や|ゅ|ゆ|ょ|よ|ら|り|る|れ|ろ|ゎ|わ|ゐ|ゑ|を|ん|ゔ|ゕ|ゖ|ゝ|ゞ|ゟ|゠|ァ|ア|ィ|イ|ゥ|ウ|ェ|エ|ォ|オ|カ|ガ|キ|ギ|ク|グ|ケ|ゲ|コ|ゴ|サ|ザ|シ|ジ|ス|ズ|セ|ゼ|ソ|ゾ|タ|ダ|チ|ヂ|ッ|ツ|ヅ|テ|デ|ト|ド|ナ|ニ|ヌ|ネ|ノ|ハ|バ|パ|ヒ|ビ|ピ|フ|ブ|プ|ヘ|ベ|ペ|ホ|ボ|ポ|マ|ミ|ム|メ|モ|ャ|ヤ|ュ|ユ|ョ|ヨ|ラ|リ|ル|レ|ロ|ヮ|ワ|ヰ|ヱ|ヲ|ン|ヴ|ヵ|ヶ|ヷ|ヸ|ヹ|ヺ]+')
+
+title = "VITS"
+description = "demo for VITS: Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech. To use it, simply add your text, or click one of the examples to load them. Read more at the links below."
+article = "<p style='text-align: center'><a href='https://arxiv.org/abs/2106.06103'>Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech</a> | <a href='https://github.com/jaywalnut310/vits'>Github Repo</a></p>"
+
+examples = [
+["原因不明の海面上昇によって、地表の多くが海に沈んだ近未来。"],
+["幼い頃の事故によって片足を失った少年・斑鳩夏生は、都市での暮らしに見切りを付け、海辺の田舎町へと移り住んだ。"],
+["身よりのない彼に遺されたのは、海洋地質学者だった祖母の船と潜水艇、そして借金。"],
+["nanika acltara itsudemo hanashIte kudasai. gakuiNno kotojanaku, shijini kaNsuru kotodemo nanidemo."]
+]
 
 hps = utils.get_hparams_from_file("./configs/ATR.json")
+
 net_g = SynthesizerTrn(
     len(symbols),
     hps.data.filter_length // 2 + 1,
@@ -60,10 +67,21 @@ net_g = SynthesizerTrn(
     **hps.model)
 _ = net_g.eval()
 
-_ = utils.load_checkpoint("./logs/ATR/G_74000.pth", net_g, None)
+_ = utils.load_checkpoint("./G_85000.pth", net_g, None)
+
+
+def get_text(text, hps):
+    text_norm = cleaned_text_to_sequence(text)
+    if hps.data.add_blank:
+        text_norm = commons.intersperse(text_norm, 0)
+    text_norm = torch.LongTensor(text_norm)
+    return text_norm
 
 def jtts(text):
-  stn_tst = get_text(japanese_phrase_cleaners(text), hps)
+  if jp_match.match(text):
+    stn_tst = get_text(japanese_phrase_cleaners(text), hps)
+  else:
+    stn_tst = get_text(text, hps)
   with torch.no_grad():
     x_tst = stn_tst.unsqueeze(0)
     x_tst_lengths = torch.LongTensor([stn_tst.size(0)])
@@ -71,16 +89,7 @@ def jtts(text):
     scipy.io.wavfile.write("out.wav", hps.data.sampling_rate, audio)
     return "./out.wav"
 
-inputs = gr.inputs.Textbox(lines=5, label="Input Text")
-outputs =  gr.outputs.Audio(label="Output Audio")
-
-
-title = "VITS"
-description = "demo for VITS: Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech. To use it, simply add your text, or click one of the examples to load them. Read more at the links below."
-article = "<p style='text-align: center'><a href='https://arxiv.org/abs/2106.06103'>Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech</a> | <a href='https://github.com/jaywalnut310/vits'>Github Repo</a></p>"
-
-examples = [
- ["吾輩は猫である。名前はまだない"],
- ["試験勉強頑張ってくださいね"]]
-
-gr.Interface(jtts, inputs, outputs, title=title, description=description, article=article, examples=examples).launch()
+if __name__ == '__main__':
+  inputs = gr.inputs.Textbox(lines=5, label="Input Text")
+  outputs =  gr.outputs.Audio(label="Output Audio")
+  gr.Interface(jtts, inputs, outputs, title=title, description=description, article=article, examples=examples).launch()

logs/ATR/G_74000.pth

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:03465e5b6a4de27bf15e59b714463f8cf5e188ac018f9225e5380b22645e1c2d
-size 436558785

requirements.txt

@@ -8,4 +8,6 @@ tensorboard==2.3.0
 torch==1.6.0
 torchvision==0.7.0
 Unidecode==1.1.1
-torchtext
+pyopenjtalk
+janome
+regex