app.py

import json
from textwrap import indent

import gradio as gr
import os

from humanfriendly.terminal import output

from app.modules.processing import transcript_audio, save_transcript
from app.modules.summarizer import process_transcript_file, generate_meeting_minutes
from app.config import *
from app.modules.exporter import export_meeting_minutes_to_docx


def process_audio_to_doc(audio_path: str, open_ai_key: str) -> str:
    """
    Nhận 1 file audio -> Meeting minutes
    :param audio_path:
    :return:
    """

    os.environ['OPENAI_API_KEY'] = open_ai_key

    if not audio_path:
        raise ValueError("No audio path provided")

    transcript = transcript_audio(audio_path, device=WHISPER_DEVICE, vad_filter=WHISPER_USE_VAD,
                                  beam_size=WHISPER_BEAM_SIZE, model_size=WHISPER_MODEL_SIZE)
    print(transcript)
    temp_path = 'temp_transcript.txt'

    save_transcript(transcript, temp_path)

    #Xử lý transcript
    meeting_minutes = process_transcript_file(temp_path)

    print(meeting_minutes)

    # Refine
    output_docx = "generated_meeting_minutes.docx"
    export_meeting_minutes_to_docx(meeting_minutes, output_docx)


    return output_docx


#Xây dựng 1 giao diện Gradio

iface = gr.Interface(
    fn=process_audio_to_doc,
    inputs= [gr.Audio(type="filepath", label="Tải lên file audio"), gr.Text(label="API_open_ai key")],
    outputs = gr.File(label="Tải xuống docx cuộc họp"),
    title="Meeting minutes",
    description="Taỉ audio và trả về biên bản cuộc họp",
)

if __name__ == "__main__":
    iface.launch()