app.py

import gradio as gr
import numpy as np
import tempfile
import scipy.io.wavfile as wav

# -------------------------------
# 1. Load Models (Lightweight)
# -------------------------------

from transformers import pipeline

# Speech-to-Text (Whisper)
stt = pipeline("automatic-speech-recognition", model="openai/whisper-small")

# Better Tutor Model (FLAN-T5)
llm = pipeline("text-generation", model="distilgpt2")


# -------------------------------
# 2. Core Functions
# -------------------------------

def speech_to_text(audio):
    """
    Converts speech (audio input) to text
    """
    if audio is None:
        return "No audio provided."

    sample_rate, data = audio

    with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as tmp:
        wav.write(tmp.name, sample_rate, data)
        result = stt(tmp.name)

    return result["text"]


def generate_response(text):
    if not text or text == "No audio provided.":
        return "Please provide valid input."

    prompt = f"""
    You are an AI tutor.
    Explain clearly and simply.

    Question: {text}
    Answer:
    """

    output = llm(prompt, max_length=150, num_return_sequences=1)
    response = output[0]["generated_text"]

    # Clean output
    return response.split("Answer:")[-1].strip()


# -------------------------------
# 3. Main Pipeline
# -------------------------------

def voice_tutor(audio):
    transcription = speech_to_text(audio)
    response = generate_response(transcription)

    return transcription, response


# -------------------------------
# 4. Gradio UI
# -------------------------------

with gr.Blocks() as demo:
    gr.Markdown("## 🎓 AI Voice Tutor (No TTS Version)")

    audio_input = gr.Audio(
        sources=["microphone", "upload"],
        type="numpy",
        label="Speak or Upload Audio"
    )

    transcription_box = gr.Textbox(label="Transcription")
    response_box = gr.Textbox(label="Tutor Response")

    submit_btn = gr.Button("Generate Response")

    submit_btn.click(
        fn=voice_tutor,
        inputs=audio_input,
        outputs=[transcription_box, response_box]
    )


# -------------------------------
# 5. Launch
# -------------------------------

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