audionar

app.py

@@ -11,29 +11,16 @@ from transformers import Wav2Vec2Processor
 from transformers.models.wav2vec2.modeling_wav2vec2 import Wav2Vec2Model
 from transformers.models.wav2vec2.modeling_wav2vec2 import Wav2Vec2PreTrainedModel
 import audiofile
-import unicodedata
-import textwrap
 from tts import StyleTTS2
 import audresample
-
-# -- 
-# -*- coding: utf-8 -*-
-
-# https://huggingface.co/spaces/dpc/mmstts/tree/main
-# https://huggingface.co/spaces/mms-meta/MMS/blob/main/tts.py
-
 import json
-import soundfile
 import re
 import unicodedata
-import gradio as gr
 import textwrap
-import numpy as np
-import torch
 import nltk
 from num2words import num2words
 from num2word_greek.numbers2words import convert_numbers
-from vits import VitsModel, VitsTokenizer
+from audionar import VitsModel, VitsTokenizer
 
 nltk.download('punkt', download_dir='./')
 nltk.download('punkt_tab', download_dir='./')
@@ -534,14 +521,14 @@ def audionar_tts(text=None,
                       lang_code=lang_code,
                       )[0, :]
             total_audio.append(x)
-            
+
         print(f'\n\n_______________________________ {_t} {x.shape=}')
 
     x = torch.cat(total_audio).cpu().numpy()
-    
+
     tmp_file = f'_speech.wav'
-    
-    soundfile.write(tmp_file, x, 16000)
+
+    audiofile.write(tmp_file, x, 16000)
 
     return tmp_file