app.py

import numpy as np
import gradio as gr
from gradio.mix import Series
import os 

os.environ['CUDA_LAUNCH_BLOCKING'] = '1' # added after repeated runtime erros (seems with whisper)

description = "This space mixes the large-multilingual [emotion classification model of DReAMy](https://github.com/lorenzoscottb/DReAMy) with the whisper model from OpenAI, to directly classify recording according to the Hall and Van de Castle framework, with respect to emotions."
title       = "DReAM v. II"

whisper = gr.Interface.load("models/openai/whisper-base")

interface_model_L = gr.Interface.load("models/DReAMy-lib/xlm-roberta-large-DreamBank-emotion-presence")

Series(
    whisper, 
    interface_model_L,
    description = description,
    title = title,
    inputs = gr.Audio(source="microphone"),
).launch()