Spaces:
Runtime error
Runtime error
Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import streamlit as st
|
| 2 |
+
import tempfile
|
| 3 |
+
import os
|
| 4 |
+
from TTS.config import load_config
|
| 5 |
+
from TTS.utils.manage import ModelManager
|
| 6 |
+
from TTS.utils.synthesizer import Synthesizer
|
| 7 |
+
from TTS.utils.download import download_url
|
| 8 |
+
|
| 9 |
+
# Define constants
|
| 10 |
+
MAX_TXT_LEN = 800
|
| 11 |
+
MODEL_INFO = [
|
| 12 |
+
# ["Model Name", "Model File", "Config File", "URL"]
|
| 13 |
+
# Add other models in the same format
|
| 14 |
+
["vits-espeak-57000", "checkpoint_57000.pth", "config.json", "https://huggingface.co/mhrahmani/persian-tts-vits-0/tree/main"],
|
| 15 |
+
# ...
|
| 16 |
+
]
|
| 17 |
+
|
| 18 |
+
# Download models
|
| 19 |
+
def download_models():
|
| 20 |
+
for model_name, model_file, config_file, url in MODEL_INFO:
|
| 21 |
+
directory = model_name
|
| 22 |
+
os.makedirs(directory, exist_ok=True)
|
| 23 |
+
download_url(f"{url}{model_file}", directory, str(model_file))
|
| 24 |
+
download_url(f"{url}{config_file}", directory, "config.json")
|
| 25 |
+
|
| 26 |
+
# Load a model and perform TTS
|
| 27 |
+
def synthesize_speech(text, model_name):
|
| 28 |
+
if len(text) > MAX_TXT_LEN:
|
| 29 |
+
text = text[:MAX_TXT_LEN]
|
| 30 |
+
st.warning(f"Input text was truncated to {MAX_TXT_LEN} characters.")
|
| 31 |
+
|
| 32 |
+
synthesizer = Synthesizer(f"{model_name}/best_model.pth", f"{model_name}/config.json")
|
| 33 |
+
if synthesizer is None:
|
| 34 |
+
st.error("Model not found!")
|
| 35 |
+
return None
|
| 36 |
+
|
| 37 |
+
wavs = synthesizer.tts(text)
|
| 38 |
+
with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as fp:
|
| 39 |
+
synthesizer.save_wav(wavs, fp)
|
| 40 |
+
return fp.name
|
| 41 |
+
|
| 42 |
+
# Streamlit app
|
| 43 |
+
def main():
|
| 44 |
+
st.title('persian tts playground')
|
| 45 |
+
st.markdown("""
|
| 46 |
+
Persian TTS Demo)
|
| 47 |
+
""")
|
| 48 |
+
|
| 49 |
+
text_input = st.text_area("Enter Text to Synthesize:", "زین همرهان سست عناصر، دلم گرفت.")
|
| 50 |
+
model_name = st.selectbox("Pick a TTS Model", [info[0] for info in MODEL_INFO], index=1)
|
| 51 |
+
|
| 52 |
+
if st.button('Synthesize'):
|
| 53 |
+
audio_file = synthesize_speech(text_input, model_name)
|
| 54 |
+
if audio_file:
|
| 55 |
+
st.audio(audio_file, format='audio/wav')
|
| 56 |
+
|
| 57 |
+
# Download models and run the Streamlit app
|
| 58 |
+
if __name__ == "__main__":
|
| 59 |
+
download_models()
|
| 60 |
+
main()
|