app.py

import gradio as gr

from transformers import pipeline

s2t=gr.Interface.load('huggingface/facebook/s2t-medium-librispeech-asr')

grammar = gr.Interface.load('huggingface/prithivida/grammar_error_correcter_v1')


def out(audio1,audio2):
  if (audio1==None) and (audio2==None):
    return "no audio","no audio"
  elif audio1==None:

    x=s2t(audio2)
    return x, grammar(x)
  else:
    x=s2t(audio1)
    return x, grammar(x)

iface = gr.Interface(
  fn=out, 
  title="Speech Audio to text (with corrected grammar)",
  description="2 possible input methods of audio file, transformed to text (as an output) and corrected grammar after(another output)!",
  inputs=[gr.inputs.Audio(source="upload", type="filepath", label=None, optional=True),                                           
          gr.inputs.Audio(source="microphone", type="filepath", label=None, optional=True)],
  examples=[["Grammar-Correct-Sample.mp3"], ["Grammar-Wrong-Sample.mp3"],],
  outputs=['text','text']
)

iface.launch(enable_queue=True,show_error=True)