app.py

import streamlit as st
import langcodes
from allosaurus.app import read_recognizer

  



def get_langcode_for_allosaurus(input_code):
  langcode = "ipa" # the default allosaurus recognizer
  try:
    lang = langcodes.get(input_code)
    langcode = lang.to_alpha3()
  except langcodes.LanguageTagError as e:
    pass
  return langcode 
  
  
  


if __name__ == "__main__":
  input_code = st.text_input("(optional) 2 or 3-letter ISO code for input language", max_chars=3)
  langcode = get_langcode_for_allosaurus(input_code)
  st.write(f"Instructing allosaurus to recognize using language {langcode}")
  
  model = read_recognizer()
  
  uploaded_file = st.file_uploader("Choose a file")
  if uploaded_file is not None:
    audio_bytes = uploaded_file.getvalue()
#    audio_bytes = audio_file.read()
    st.audio(audio_bytes, format='audio/wav')
    
    result = model.recognize(audio_bytes, langcode)
    st.write(result)