app.py

import os
import subprocess
import openai
import gradio as gr

openai.api_key = "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"

def transcribe(audio):
    with open(audio, "rb") as audio_file:
        transcript = openai.Audio.transcribe("whisper-1", audio_file)
    return transcript["text"]

def generate_response(transcribed_text):
    response = openai.ChatCompletion.create(
        model="gpt-3.5-turbo",
        messages=[
            {"role": "system", "content": "Wewe ni mtaalamu wa Kilimo cha mazao, na kilimo cha viazi lishe utajibu maswali yote kwa kiswahili"},
            {"role": "user", "content": "nini maana ya Viazi lishe?"},
            {"role": "assistant", "content": """ viazi lishe ni Viazi vitamu vyenye rangi ya karoti kwa ndani ambavyo vina vitamin A kwa wingi"""},
            {"role": "user", "content": "nini matumizi ya viazi lishe?"},
            {"role": "assistant", "content": """ viazi lishe vinaweza kutengenezea chakula kama Keki, 
            Maandazi, Kalimati na tambi: Ukisaga unga wa viazi lishe, 
            unaweza kutumika kupika vyakula ambavyo huwa watu 
            hutumia unga wa ngano kupika, unga wa viazi lishe una 
            virutubisho vingi zaidi kuliko unga wa ngano na 
            ukitumika kupikia vyakula tajwa hapo juu watumiaji 
            watakuwa wanakula vyakula vyenye virutubisho Zaidi."""},
            {"role": "user", "content": transcribed_text},
        ]
    )
    return response['choices'][0]['message']['content']

def inference(text):
    output_file = "tts_output.wav"
    cmd = ['tts', '--text', text, '--out_path', output_file]
    subprocess.run(cmd, check=True)
    return output_file

def process_audio_and_respond(audio):
    text = transcribe(audio)
    response_text = generate_response(text)
    output_file = inference(response_text)
    return response_text, output_file

demo = gr.Interface(
    process_audio_and_respond,
    gr.inputs.Audio(source="microphone", type="filepath", label="Speak your question"),
    [gr.outputs.Textbox(label="Answer (text)"), gr.outputs.Audio(type="filepath", label="Answer (audio)")],
    title="AI Question Answering",
    description="Ask any question and get an AI-generated answer as audio output.",
    theme="compact",
    layout="vertical",
    allow_flagging=False,
    live=True,
)

demo.launch()