Rajiラジ
Back to projects
Case Studylocal-firstTypeScriptarchitecture

Example Project

This is going to be random AI generated gibberish to test the content renderer and the MDX parser. Don't actually read this or take it seriously. This will soon be gone, along with all the dummy contents.

ExampleAppName

ExampleAppName is an experimental vocal-analysis project exploring whether useful information about vocal resonance can be extracted from ordinary audio input.

The goal isn't to build a perfect scientific instrument. The goal is to build something that makes vocal practice more observable.

A singer can hear that something changed.

ExampleAppName asks:

Can we also make that change visible?

The Problem

Vocal practice contains an uncomfortable amount of ambiguity.

You can change your vowel shape, resonance placement, breath coordination, or vocal tract configuration and immediately feel that something is different. But the feedback loop is usually subjective:

  • Did the sound actually become brighter?
  • Did the resonance shift?
  • Was the improvement consistent?
  • Was it the vowel, the pitch, the volume, or something else?
  • Can the same sensation be reproduced tomorrow?

For someone practicing seriously, better feedback can be more useful than more information.

ExampleAppName is an attempt to explore that feedback loop.

What Makes This Interesting

Audio is already full of measurable properties.

A signal contains information about:

  • amplitude
  • frequency
  • harmonic structure
  • spectral distribution
  • temporal changes
  • noise
  • periodicity

The interesting question is not whether these properties exist.

The interesting question is which of them are useful to a singer.


The Approach

The first version deliberately avoids trying to solve everything.

Instead, it focuses on a small collection of measurements that can be calculated from incoming audio and presented in a way that can be interpreted quickly.

The general pipeline looks roughly like this:

Microphone

Audio input

Signal preprocessing

Feature extraction

Temporal smoothing

Visual feedback