app.py

"""
Gradio note-taking app that:
1. Records voice via microphone
2. Transcribes to text with Whisper (openai/whisper-large-v3)
3. Generates a diagram image from the text with FLUX
4. Displays the note and the diagram side-by-side
"""

import os
import tempfile
import gradio as gr
from huggingface_hub import InferenceClient

# ------------------------------------------------------------------
# Configuration
# ------------------------------------------------------------------
HF_TOKEN = os.getenv("HF_TOKEN")  # export HF_TOKEN=...
if not HF_TOKEN:
    raise RuntimeError("Set HF_TOKEN environment variable")

client = InferenceClient(
    provider="fal-ai",
    api_key=HF_TOKEN,
    bill_to="huggingface",
)

# ------------------------------------------------------------------
# Core helpers
# ------------------------------------------------------------------


def transcribe(audio_path: str) -> str:
    """Transcribe audio file to text using Whisper."""
    transcription = client.automatic_speech_recognition(
        audio_path,
        model="openai/whisper-large-v3",
    )
    return transcription["text"]


def generate_diagram(text: str) -> str:
    """Generate a diagram image from text using FLUX, save to tmp file and return path."""
    image = client.text_to_image(
        prompt=f"Clean, simple diagram illustrating: {text}",
        model="black-forest-labs/FLUX.1-schnell",
        width=768,
        height=512,
    )

    tmp = tempfile.NamedTemporaryFile(suffix=".png", delete=False)
    image.save(tmp.name)
    return tmp.name


# ------------------------------------------------------------------
# Gradio UI
# ------------------------------------------------------------------

def process_voice(audio):
    """Chain transcription + diagram generation."""
    text = transcribe(audio)
    img_path = generate_diagram(text)
    return text, img_path


with gr.Blocks(title="Voice-to-Diagram Note Taker") as demo:
    gr.Markdown("# 🎤 Voice Note & Diagram Generator")
    gr.Markdown("Speak into the microphone; your words become a note and an auto-generated diagram.")

    with gr.Row():
        mic = gr.Audio(sources="microphone", type="filepath", label="Record")
    with gr.Row():
        with gr.Column(scale=2):
            note_text = gr.Textbox(label="Transcription", lines=5, interactive=True)
        with gr.Column(scale=1):
            diagram_img = gr.Image(label="Generated Diagram")

    mic.change(fn=process_voice, inputs=mic, outputs=[note_text, diagram_img])

demo.launch()