Initial commit

app.py

@@ -0,0 +1,83 @@
+from TTS.api import TTS
+from bs4 import BeautifulSoup
+import requests
+import streamlit as st
+import tempfile
+import os
+import json
+
+with open('config.json', 'r') as f:
+    config = json.load(f)
+
+LANGUAGES_URL = config['LANGUAGES_URL']
+OUTPUT_FILENAME = config['OUTPUT_FILENAME']
+
+def get_iso_languages():
+    response = requests.get(LANGUAGES_URL)
+    soup = BeautifulSoup(response.text, 'html.parser')
+
+    p_tags = soup.find_all('p')
+
+    iso_language_dict = {}
+
+    for p_tag in p_tags[1:]:  # Skipping the first <p> which contains the header
+        parts = p_tag.get_text().split()
+        if len(parts) == 2:
+            iso_code, language_name = parts
+            iso_language_dict[language_name] = iso_code
+
+    return iso_language_dict
+
+def create_temp_file(input_wav):
+    temp_file = tempfile.NamedTemporaryFile(delete=False)
+    temp_file.write(input_wav.read())
+    return temp_file
+
+def remove_temp_file(temp_file):
+    temp_file.close()
+    os.remove(temp_file.name)
+
+def update_progress(percent, text):
+    progress_bar.progress(percent)
+    status_text.text(text)
+
+iso_languages = get_iso_languages()
+languages = list(iso_languages.keys())
+
+language = st.selectbox('Select a language', languages)
+text = st.text_input('Enter some text')
+input_wav = st.file_uploader("Upload a WAV file", type=["wav"])
+
+if input_wav:
+    if not input_wav or input_wav is None:
+        st.error('Please upload wav input audio')
+    elif not text:
+        st.error('Please write text')
+    else:
+        progress_bar = st.progress(0)
+        status_text = st.empty()
+
+        temp_file = create_temp_file(input_wav)
+
+        iso_code = iso_languages[language]
+
+        update_progress(0, 'Loading TTS model...')
+        api = TTS(f"tts_models/{iso_code}/fairseq/vits")
+        
+        update_progress(50, 'Generating audio...')
+        api.tts_with_vc_to_file(
+            text,
+            speaker_wav=temp_file.name,
+            file_path=OUTPUT_FILENAME
+        )
+
+        remove_temp_file(temp_file)
+
+        audio_file = open(OUTPUT_FILENAME, 'rb')
+        audio_bytes = audio_file.read()
+
+        update_progress(100, 'Audio generated successfully!')
+
+        st.audio(audio_bytes, format='audio/wav')
+
+        st.download_button('Download WAV', data=audio_bytes, file_name='file.wav')

config.json

@@ -0,0 +1,7 @@
+{
+    "APP_NAME": "TTS-Voice-Cloner",
+    "APP_LOGO": "",
+    "APP_DESCRIPTION": "The service allows you to clone a voice from a given audio recording and produce text voice acting",
+    "LANGUAGES_URL": "https://dl.fbaipublicfiles.com/mms/tts/all-tts-languages.html",
+    "OUTPUT_FILENAME": "output.wav"
+}

requirements.txt

@@ -0,0 +1,2 @@
+beautifulsoup4
+TTS