app.py

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

  
def get_path_to_wav_format(uploaded_file):
  st.write(dir(uploaded_file))
  st.write(type(uploaded_file))
  if ".wav" in uploaded_file:
    return path(uploaded_file)
  # TODO: .mp3 conversion      


def get_langcode_for_allosaurus(input_code):
  langcode = "ipa" # the default allosaurus recognizer
  description = "the default universal setting, not specific to any language"
  try:
    lang = langcodes.get(input_code)
    langcode = lang.to_alpha3()
    description = lang.display_name()
  except langcodes.LanguageTagError as e:
    pass
  return langcode, description 
  
  
  


if __name__ == "__main__":
  input_code = st.text_input("(optional) 2 or 3-letter ISO code for input language", max_chars=3)
  langcode, description = get_langcode_for_allosaurus(input_code)
  st.write(f"Instructing Allosaurus to recognize using language {langcode}. That is, {description}")
  
  model = read_recognizer()
  
  uploaded_file = st.file_uploader("Choose a file", type=[
                                                          ".wav", 
                                                          ".mp3",
                                                          ])
  
  
  if uploaded_file is not None:

    
    
#    audio_bytes = audio_file.read()
    st.audio(uploaded_file, format='audio/wav')
    
    wav_file = get_path_to_wav_format(uploaded_file)    
    
    result = model.recognize(wav_file, langcode)
    st.write(result)