Spaces:
Running
Running
import gradio as gr | |
from gtts import gTTS | |
from pydub import AudioSegment | |
import numpy as np | |
import os | |
def text_to_speech(prompt): | |
# gTTS๋ฅผ ์ด์ฉํด Bulgarian ํ ์คํธ๋ฅผ ์์ฑ์ผ๋ก ๋ณํ | |
tts = gTTS(text=prompt, lang="bg") | |
audio_file = "output.mp3" | |
tts.save(audio_file) | |
# pydub๋ฅผ ์ฌ์ฉํ์ฌ mp3 ํ์ผ์ ๋ถ๋ฌ์ต๋๋ค. | |
sound = AudioSegment.from_mp3(audio_file) | |
# pydub์ raw ๋ฐ์ดํฐ๋ฅผ numpy ๋ฐฐ์ด๋ก ๋ณํ (int16) | |
samples = np.array(sound.get_array_of_samples()) | |
# ๋ง์ฝ ์คํ ๋ ์ค๋ผ๋ฉด ๋ชจ๋ ธ๋ก ๋ณํ (์ฑ๋ ํ๊ท ) | |
if sound.channels > 1: | |
samples = samples.reshape((-1, sound.channels)) | |
samples = samples.mean(axis=1) | |
# int16 ๋ฐ์ดํฐ๋ฅผ float32๋ก ์ ๊ทํ (๋ฒ์: [-1.0, 1.0]) | |
samples = samples.astype(np.float32) / 32768.0 | |
sample_rate = sound.frame_rate | |
# ์์๋ก ์์ฑํ mp3 ํ์ผ ์ญ์ | |
os.remove(audio_file) | |
# gr.Audio(type="numpy")๋ (numpy_array, sample_rate) ํํ์ ๊ธฐ๋ํฉ๋๋ค. | |
return samples, sample_rate | |
with gr.Blocks() as demo: | |
gr.Markdown("## Bulgarian Text-to-Speech (TTS)") | |
with gr.Row(): | |
input_prompt = gr.Textbox(label="Enter a prompt in Bulgarian:") | |
# type์ "numpy"๋ก ์ค์ ํ์ฌ numpy ๋ฐฐ์ด์ ์ฌ์ฉํฉ๋๋ค. | |
output_audio = gr.Audio(label="Generated Speech", type="numpy") | |
generate_button = gr.Button("Generate Speech") | |
generate_button.click(text_to_speech, inputs=input_prompt, outputs=output_audio) | |
if __name__ == "__main__": | |
demo.launch() | |