app.py

import gradio as gr
import os
from TTS.api import TTS
import time


"""
For Fairseq models, use the following name format: tts_models/<lang-iso_code>/fairseq/vits. You can find the language ISO codes here and learn about the Fairseq models here.
https://dl.fbaipublicfiles.com/mms/tts/all-tts-languages.html
"""

api_rus = TTS(f"tts_models/rus/fairseq/vits", gpu=False)
api_eng = TTS(f"tts_models/eng/fairseq/vits", gpu=False)
api_spa = TTS(f"tts_models/spa/fairseq/vits", gpu=False)
api_fas = TTS(f"tts_models/fas/fairseq/vits", gpu=False)  # Persian
api_tur = TTS(f"tts_models/tur/fairseq/vits", gpu=False)  # Turkish
api_deu = TTS(f"tts_models/deu/fairseq/vits", gpu=False)  # German, Standard
api_ara = TTS(f"tts_models/ara/fairseq/vits", gpu=False)  # Arabic
api_por = TTS(f"tts_models/por/fairseq/vits", gpu=False)  # Portuguese
api_hin = TTS(f"tts_models/hin/fairseq/vits", gpu=False)  # Hindi


def audio_tts(txt, language, audio_file):

    # TTS with on the fly voice conversion
    print(f"Language: {language}")
    if language == "rus":
        api_rus.tts_with_vc_to_file(txt, speaker_wav=audio_file, file_path="ouptut.wav")
    elif language == "eng":
        api_eng.tts_with_vc_to_file(txt, speaker_wav=audio_file, file_path="ouptut.wav")
    elif language == "spa":
        api_spa.tts_with_vc_to_file(txt, speaker_wav=audio_file, file_path="ouptut.wav")
    elif language == "fas":
        api_fas.tts_with_vc_to_file(txt, speaker_wav=audio_file, file_path="ouptut.wav")
    elif language == "tur":
        api_tur.tts_with_vc_to_file(txt, speaker_wav=audio_file, file_path="ouptut.wav")
    elif language == "deu":
        api_deu.tts_with_vc_to_file(txt, speaker_wav=audio_file, file_path="ouptut.wav")
    elif language == "ara":
        api_ara.tts_with_vc_to_file(txt, speaker_wav=audio_file, file_path="ouptut.wav")
    elif language == "por":
        api_por.tts_with_vc_to_file(txt, speaker_wav=audio_file, file_path="ouptut.wav")
    elif language == "hin":
        api_hin.tts_with_vc_to_file(txt, speaker_wav=audio_file, file_path="ouptut.wav")
    return "ouptut.wav"
    

demo = gr.Interface(fn=audio_tts, inputs=[gr.Textbox(label="Input text TTS", value="Привет! Я Макс."),
                                          gr.Dropdown(choices=["rus", "eng", "spa", "fas", "tur", "deu", "ara", "por", "hin"], 
                                                      label="Language", value="rus"),
                                          gr.Audio(source="upload", type="filepath", label="Input audio")],
                    outputs=gr.Audio(source="upload", type="filepath", label="Output audio"))

demo.queue(concurrency_count=1).launch(show_error=True)