app.py

# https://huggingface.co/spaces/lampongyuen/Gradio-Webcam-Streaming/blob/main/app.py  (I thought needed video instead of Image)

import gradio as gr
import numpy as np 
import time

def flip(im):
    return np.flipud(im)

def dummyaudioprocess(aud):
    time.sleep(3)
    return aud

with gr.Blocks() as realtimemedia:
    gr.HTML("Dual Audio and Image and processing test - With inspiration from - https://huggingface.co/spaces/lampongyuen/Gradio-Webcam-Streaming/blob/main/app.py (I thought needed video instead of Image) <br> End goal is SAM on live webcam + audio live with whisper <br> Live interface seems not to work in colab so dont debug there (doesnt work for me here, why?)" )
    gr.Interface(flip, gr.Image(sources=["webcam"], streaming=True), "image", live=True)
    gr.Interface(dummyaudioprocess, gr.Audio(sources=["microphone"], streaming=True), "audio") #, live=True)

realtimemedia.queue().launch()