app.py

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import os
import librosa
import joblib
import gradio as gr

## Voice Data Feature Extraction

### extract the features from the audio files using mfcc
def feature_extracter(fileName):
    audio,sample_rate = librosa.load(fileName,sr=None, mono=True,  dtype=np.float32,res_type='kaiser_fast')
    mfcc_features = librosa.feature.mfcc(y=audio,sr=sample_rate,n_mfcc=30)
    mfccs_scaled_features = np.mean(mfcc_features.T, axis=0)
    
    return list(mfccs_scaled_features)

def prediction_age_gender(fileName):

  col_name = ['Feature_1', 'Feature_2', 'Feature_3', 'Feature_4', 'Feature_5','Feature_6', 'Feature_7', 'Feature_8', 'Feature_9', 'Feature_10','Feature_11', 'Feature_12', 'Feature_13', 'Feature_14', 'Feature_15','Feature_16', 'Feature_17', 'Feature_18', 'Feature_19', 'Feature_20','Feature_21', 'Feature_22', 'Feature_23', 'Feature_24', 'Feature_25','Feature_26', 'Feature_27', 'Feature_28', 'Feature_29', 'Feature_30']

  observation = [feature_extracter(fileName)]
  observation = pd.DataFrame(observation, columns = col_name)

  ## scaling the observation
  scaler = joblib.load('/content/scaler.pkl')
  scaled_observation = scaler.transform(observation)
  scaled_observation = pd.DataFrame(scaled_observation, columns = col_name)  

  ### Gender classification model
  gender_model = joblib.load('/content/KNN_gender_detection.pkl')
  gender_predict = gender_model.predict_proba(scaled_observation)
  ## considering the labels 1 = male 0 = female
  gender_dict = {}
  gender_dict['Female'] = gender_predict[0][0]
  gender_dict['Male'] = gender_predict[0][1]

  ### Age classification model
  age_model = joblib.load('/content/KNN_age_model.pkl')
  age_predict = age_model.predict_proba(scaled_observation)
  age_dict = {}
  age_dict['Eighties'] = age_predict[0][0]
  age_dict['Fifties'] = age_predict[0][1]
  age_dict['Fourties'] = age_predict[0][2]
  age_dict['Seventies'] = age_predict[0][3]
  age_dict['Sixties'] = age_predict[0][4]
  age_dict['Teens'] = age_predict[0][5]
  age_dict['Thirties'] = age_predict[0][6]
  age_dict['Twenties'] = age_predict[0][7]
  age_dict['Other'] = 1 - age_dict['Eighties'] - age_dict['Fifties'] - age_dict['Fourties'] - age_dict['Seventies'] - age_dict['Sixties']  - age_dict['Teens'] - age_dict['Thirties'] - age_dict['Twenties']

  #final = "The person is a: " + gender + " of the age group: " + age
  return gender_dict, age_dict

demo = gr.Interface(
    prediction_age_gender,
    inputs = [gr.Audio(sources=["microphone","upload"], type = 'filepath')],
    outputs = [gr.Label(num_top_classes=2, label = 'Gender'), gr.Label(num_top_classes=9, label = 'Age Class')],
    #gr.Text()
).launch(share=True, debug = True)