Ever wonder who do you or have the potential to recite like. This AI powered web app made using Teachable Machine might help you. Trained on recitation data of Sheikh Abdul Basit Abdul Samad, Mishary bin Rashid, Abu Bakr al-Shatri, Abdur-Rahman as-Sudais, Junaid Jamshed, and Omer Amin. It detects the similarities between your and the reciter's voice using CNNs. This App is by no means accurate and is at a very early prototype stage. It should be noted that this App is far from perfect and implementation is questionable. I failed to achieve what I wanted (Sort of).

What I wanted vs. what I was able to do.

At first, the plan was to train the model using the Tensorflow python api through Google Colab, and then implement the trained model and architecture on web though Tensorflow.js but I failed to achieve it due to my lack of experience in JavaScript, and tight schedule. At the end, I had to cheat and use Teachable Machine and ml5.js to create and train a model, but that solution is less flexible. I also think implementation and use of CNNs is questionable for this task, but again it caan be due to a small dataset. In the future, I will research newer ways to make this web app better. All this, it was still a joy to participate in my first hackathon, and I hope that the Ummahacks team continue to conduct such events in the future, Insha'Allah.

HTML5 Template by HTML5 UP.

Built With

Share this project: