gistfile1.txt

# pip3 install git+https://github.com/espnet/espnet
# pip3 install "espnet[all]" <--- deps, not sure if espnet needs to be reinstalled from source after this, just make sure its March1 or later release
# pip3 install pyopenjtalk

import pyaudio
from espnet2.bin.tts_inference import Text2Speech
import numpy as np

import soundfile as sf

from pydub import AudioSegment, playback

# Initialize the pre-trained TTS model
text2speech = Text2Speech.from_pretrained("espnet/kan-bayashi_tsukuyomi_full_band_vits_prosody")

# Set the audio sample rate to 24000Hz
sample_rate = 24000

while True:
    # Take user input for Japanese text to generate speech from
    text_input = input("Enter Japanese text to generate speech: ")
    # Generate speech from the input text using the pre-trained TTS model
    wav = text2speech(text_input)["wav"]
    # Save
    sf.write("/tmp/out.wav", wav.numpy(), text2speech.fs, "PCM_16")
    # Play
    audio_file = AudioSegment.from_file("/tmp/out.wav", format="wav")
    playback.play(audio_file)

AbeShinzo0708/ESPnet_VITS_SugaYoshihide

JP prime minister