Spaces:
Running
Running
import gradio as gr | |
from gtts import gTTS | |
from pydub import AudioSegment | |
import numpy as np | |
import os | |
import io | |
def text_to_speech(prompt): | |
# gTTS๋ฅผ ์ฌ์ฉํ์ฌ Bulgarian ํ ์คํธ๋ฅผ ์์ฑ์ผ๋ก ๋ณํ | |
tts = gTTS(text=prompt, lang="bg") | |
audio_file = "output.mp3" | |
tts.save(audio_file) | |
# pydub์ ์ฌ์ฉํ์ฌ mp3 ํ์ผ์ ๋ก๋ํฉ๋๋ค. | |
sound = AudioSegment.from_mp3(audio_file) | |
# pydub์ raw data๋ฅผ numpy ๋ฐฐ์ด๋ก ๋ณํ (16๋นํธ ์ ์ํ) | |
samples = np.array(sound.get_array_of_samples()) | |
# ๋ชจ๋ ธ ์ฑ๋์ด ์๋๋ผ๋ฉด, ์ฑ๋์ ํฉ์นฉ๋๋ค. | |
if sound.channels > 1: | |
samples = samples.reshape((-1, sound.channels)) | |
samples = samples.mean(axis=1).astype(np.int16) | |
sample_rate = sound.frame_rate | |
# ์์ 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:") | |
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() | |