UrbanSound8k

Project is live at : UrbanSound8k

Dockerized installation in local machine

• paste below code in cmd

  git clone https://github.com/BhavyBansal24/UrbanSound8k.git
  docker build -t ultra_sound8k .
  docker run -dp 8501:8501 --name SoundDetect ultra_sound8k
  timeout 2 /nobreak && start http://localhost:8501
  
  

How to use UrbanSound8k web-app:

• Click here
• Click on Browse Files
• Upload a (.wav) audio file, Model will classify your uploaded sound from the classes.
• classes are : ['dog_bark' , 'children_playing' , 'car_horn' , 'air_conditioner' , 'street_music' , 'gun_shot' , 'siren' , 'engine_idling' , 'jackhammer' , 'drilling']
• After uploading completes, you can see the prediction on right side as shown below
• Below that you can also view other analysis such as spectrogram, Harmonic Percussive Separation, Mel-Frequency Cepstral Coefficients (MFCC), Zero Crossing Rate and many more.

---

Model Details :

• Sequential model
• Training of 15 Epoch

---

Model Results :

• Model Accuracy
  • 
• Model Loss
  • 

---

Model Evaluation

• Classification Report
  • 
• Confusion Matrix
  • 

---

Dataset used for Training model:

• Dataset is taken from kaggle and link for dataset is here
• This dataset contains 8732 labeled sound of urban sounds from 10 classes: air_conditioner, car_horn, children_playing, dog_bark, drilling, enginge_idling, gun_shot, jackhammer, siren, and street_music

---

Libraries and framework used in project :

• Pandas
• Librosa
• Numpy
• Seaborn
• Sklearn
• Matplotlib
• Tensorflow
• Keras
• Ipython
• OpenCV

---

Do check my Kaggle (ipynb)Notebook:

• Link to my kaggle notebook is here
• Do upvote notebook, Hope you like it.
• and do Not forgot to check my other notebooks on kaggle as well

---

Do 🌟 this Github Repository, Hope you have loved my work

---

Moreover don't forget to follow me on :

• github
• Kaggle
• Linkedin
• Twitter
• Instagram

---

Portfolio Website :

• Hope you love to know more about me, check my Portfolio Website here