Songwriting Lyric App Software Development with Word embedding-based Search Engine

by Kisko Apeles

One of my deepest passions is songwriting, as it enables people to share their thoughts and stories in a deeply empathetic and artistic way.

What songwriters need is a place where they can write down their lines and thoughts. There are lots of note writing apps available in the market. However, one of main issues with note writing apps like Google Keep, Apple Notes, and our fairly old but loved, Windows Notepad, is that we are only able to search old writings using exact match.

Oftentimes, we don’t remember the exact note, line, or thought we have written already in our note writing apps and when we do search what we have in mind, they only provide us the exact match. This makes us miss finding notes that are fairly similar to what we have in mind.

Given that, I built an AWS S3 and EC2/Lightsail-hosted lyric app (on iOS and web platforms) for the purpose of finding lyrics that I have written before, even if they are not an exact match, so I can remember and pull out these previously written lyrics and build on them instead of starting from scratch. Here’s a video of what it can do using natural language processing, specifically word embeddings:

To further aid my songwriting progress and improvement, I have also included the following features in the lyric app:

• Dashboard that counts how many lines, outlines, and songs I have drafted (as well as my daily line output counts)
  
  
• Video recording on lyrics page (to easily document your melodies in the process)

• Video recording on web browser (to search existing songs, learn from them, and practice song covers)

• Tuner / Note Identifier (to practice ear training and identify current vocal range for songwriting purposes)

So far, this lyric app has been very helpful in improving the documentation of my song drafts and accelerate the songwriting process.