app.py

from TTS.api import TTS
import streamlit as st
import time
import os

os.environ['COQUI_TOS_AGREED'] = '1'
os.environ['BARK_TOS_AGREED'] = '1'


st.header("Bas Konuş")
# Initialize TTS object when the page is loaded

@st.cache_resource()
def model_coqui():
    device = "cpu"
    tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to(device)
    return tts  # return the tts object

# @st.cache_resource()
def model_bark():
    device = "cpu"
    tts = TTS("tts_models/multilingual/multi-dataset/bark").to(device)
    return tts


col1, col2 = st.columns(2)

with col1:
    sentence = st.text_area("Konuşmak istediğiniz metni girin")

with col2:
    language = st.selectbox("Dil Seçin", ["Ceylin", "Funda"])
    if language == "Ceylin":
        speaker_wav = "sounds/glinda.mp3"
    else:
        speaker_wav = "sounds/alp.mp3"

speaker=st.button("Seslendir")




if speaker:

    col11, col22 = st.columns(2)

    with col11:
        tts_coqui = model_coqui()
        start_time_coqui = time.time()  # Start the timer
        output_path_coqui = tts_coqui.tts_to_file(text=sentence, speaker_wav=speaker_wav ,language="tr")
        end_time = time.time()  # End the timer

        execution_time = round(end_time - start_time_coqui,2)  # Calculate the time taken
        st.write(f"Coqui için tamamlanma süresi: {execution_time} saniye")  # Display the time taken

        st.audio(output_path_coqui, format="audio/mp3")

    with col22:
        tts_bark = model_bark()
        speaker_wav_2 = "speaker_embeddings/tr_speaker_0"

        start_time_bark = time.time()
        output_path_bark = tts_bark.tts_to_file(text=sentence, speaker_wav=speaker_wav_2)
        end_time = time.time()

        execution_time = round(end_time - start_time_bark,2)
        st.write(f"Bark içi tamamlanma süresi: {execution_time} saniye")

        st.audio(output_path_bark, format="audio/mp3")