app.py

import gradio as gr
import soundfile as sf
import datetime
import numpy as np


# 錄音函數
def save_audio(audio, label, sample_rate, bit_depth):
    # 產生檔案名稱
    filename = f"{datetime.datetime.now().strftime('%Y%m%d_%H%M%S')}_{label}_{sample_rate}Hz_{bit_depth}bit.wav"

    # 將音頻數據轉換成 numpy 陣列
    audio_data = np.array(audio[1])

    # 將錄音儲存為 wav 檔案
    sf.write(filename, audio_data, int(sample_rate), subtype=f'PCM_{bit_depth}')

    return filename


# 介面設計
def create_interface():
    sample_rate = gr.Dropdown(["44100", "48000"], label="選擇採樣率", value="44100")
    bit_depth = gr.Dropdown(["16", "24"], label="選擇位深度", value="16")
    label = gr.Textbox(label="輸入標記")
    audio = gr.Audio(sources="microphone", type="numpy", label="錄音", show_download_button=True,)

    interface = gr.Interface(
        fn=save_audio,
        inputs=[audio, label, sample_rate, bit_depth],
        outputs=["text"],
        title="錄音程式",
        description="錄音最大長度為 1 秒，儲存為 WAV 格式。",
    )
    return interface


# 啟動介面
if __name__ == "__main__":
    interface = create_interface()
    interface.launch()