app.py

# import library
import gradio as gr
import librosa
import pandas as pd
import numpy as np
import pickle
import os

import tensorflow as tf
from tensorflow.keras.layers.experimental import preprocessing
from tensorflow.keras.preprocessing.image import load_img, img_to_array
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Dropout, Flatten, Conv2D, MaxPool2D, BatchNormalization, Input

def get_waveform_label(file):
    lab = tf.strings.split(file, os.path.sep)[-2]
    audio_binary = tf.io.read_file(file)
    audio, _ = tf.audio.decode_wav(audio_binary)
    waveform=tf.squeeze(audio, axis=-1)
    return waveform

def get_spectrogram_label(audio):
    padding = tf.zeros([300000]-tf.shape(audio), dtype=tf.float32)
    wave = tf.cast(audio, tf.float32)
    eq_length = tf.concat([wave, padding], 0)
    spectrogram = tf.signal.stft(eq_length, frame_length=210, frame_step=110)    
    spectrogram = tf.abs(spectrogram)
    spectrogram = tf.expand_dims(spectrogram, -1)
    return spectrogram
    
 # %load saved model
model = pickle.load(open('audio_classifier_model.pkl', 'rb'))
    
def get_audio(audio):
  audio_waveform = get_waveform_label(audio)
  audio_spect = get_spectrogram_label(audio_waveform)
  final_feat = np.array([audio_spect])
  res = np.argmax(model.predict(final_feat),axis=1)
  if res == 1:
    res ="Dog Audio";
  else:
    res = "Cat Audio"
  return res
  
iface = gr.Interface(fn=get_audio,inputs=["audio"],outputs=["text"],title="CAT/DOG AUDIO CLASSIFICATION").launch(inline=False)