Substantial

Vocal Harmonizer

A compact vocal harmonization system enabling musicians to create multi-layered vocal harmonies in real-time during performances

Timeline

January 2020December 2021

Key Technologies

ElectronicsRaspberry PiESP32DSP+1 more

Background

Vocal harmonies are a highly sought after effect in the music industry. There is currently no commercially ready method that allows solo-artists to self-harmonize in real time. The goal was to create a method for all vocalists to be able to play their voice as a variety of chords using an interface that is both intuitive and non-intrusive in a live performance setting.

Objective

Create a compact vocal harmonization pedal that allowed singers deliberate control over without limiting stage presence.

Design Phase 1: Concept Generation and Wireframing

After interviewing dozens of industry professionals and musicians, I led three teammates in the creation of a competitive analysis using a "House of Quality". Using the tradeoffs and market gaps found, we designed preliminary sketches of different ways the system could function and evaluated them accordingly.
Competitive analysis and market research using House of Quality methodology to identify design requirements and tradeoffs.

Competitive analysis and market research using House of Quality methodology to identify design requirements and tradeoffs.

From this analysis, we developed six distinct concept approaches:
"All-in-one" approach, where reading the user's chord input and processing the audio would be completed on the same microcontroller held by the performer.

"All-in-one" approach, where reading the user's chord input and processing the audio would be completed on the same microcontroller held by the performer.

This design consists of an entirely software-based approach. It would have the cheapest reproduction cost, but offered the least mobility and freedom to the performer.

This design consists of an entirely software-based approach. It would have the cheapest reproduction cost, but offered the least mobility and freedom to the performer.

This concept employed the ability for performers to control the chords played from the microphone itself, not limiting their range of use to being near a computer to add effects to their vocals.

This concept employed the ability for performers to control the chords played from the microphone itself, not limiting their range of use to being near a computer to add effects to their vocals.

This concept introduced the idea of a "frequency modulation box" (FMB), eliminating the risks associated with bringing a computer to a performance. For this idea, the process of audio manipulation would be completed using only standalone devices.

This concept introduced the idea of a "frequency modulation box" (FMB), eliminating the risks associated with bringing a computer to a performance. For this idea, the process of audio manipulation would be completed using only standalone devices.

My team and I valued the benefits of the "all-in-one" approach over any of the other options, and this sketch was my primary depiction of this device. It would wrap around nearly all stage microphones using a Velcro strap and a hinge attaching the microcontrollers to the buttons for different shapes, sizes, and angles of microphone handles.

My team and I valued the benefits of the "all-in-one" approach over any of the other options, and this sketch was my primary depiction of this device. It would wrap around nearly all stage microphones using a Velcro strap and a hinge attaching the microcontrollers to the buttons for different shapes, sizes, and angles of microphone handles.

After evaluating our budget and time constraints, the team decided to utilize a disjointed approach. This allows the musician to maintain full mobility around a stage without interacting directly with a computer, while still providing room in the controller for a larger battery, circuit board, and ergonomic flexibility.

After evaluating our budget and time constraints, the team decided to utilize a disjointed approach. This allows the musician to maintain full mobility around a stage without interacting directly with a computer, while still providing room in the controller for a larger battery, circuit board, and ergonomic flexibility.

Design Phase 2: Prototyping

This flow chart depicts each step that the audio signal goes through during its conversion into a restructured chord.
Complete signal flow chart showing each step of audio processing from input to harmonized output.

Complete signal flow chart showing each step of audio processing from input to harmonized output.

As can be seen in the final sketch above, we had to create the FMB to alter the audio to a desired chord structure, as well as a remote to chose said structure.

We opted to use a Raspberry Pi 3B+ with a high fidelity hat as the FMB.

As can be seen in the final sketch above, we had to create the FMB to alter the audio to a desired chord structure, as well as a remote to chose said structure. We opted to use a Raspberry Pi 3B+ with a high fidelity hat as the FMB.

I created a custom PCB for an ESP32 microcontroller in the remote to be able to communicate with the FMB via Bluetooth.

The PCB connected the microcontroller to four buttons representing 16 combinations, a power switch, a multi-functional LED, and an internal, rechargeable battery.

I created a custom PCB for an ESP32 microcontroller in the remote to be able to communicate with the FMB via Bluetooth. The PCB connected the microcontroller to four buttons representing 16 combinations, a power switch, a multi-functional LED, and an internal, rechargeable battery.

Design Phase 3: Final Controller Design and Spectral Analysis

The final phase focused on validating our design through spectral analysis and confirming that our theoretical models matched real-world performance.
Completed vocal harmonizer system showing final controller design and FMB integration.

Completed vocal harmonizer system showing final controller design and FMB integration.

An overlay displays the idealized output that should be generated by the instrument when a user selects each chord given an input frequency of 500 Hz.

An overlay displays the idealized output that should be generated by the instrument when a user selects each chord given an input frequency of 500 Hz.

A spectrogram I designed of the actual results of each chord type produced from a recorded 500 Hz sine wave.

A spectrogram I designed of the actual results of each chord type produced from a recorded 500 Hz sine wave.

The idealized outputs and measured frequencies overlain, where one can see direct alignment.

The idealized outputs and measured frequencies overlain, where one can see direct alignment.

Key Features

  • Real-time vocal harmonization system
  • Custom PCB design for compact controller
  • Bluetooth communication between controller and processing unit
  • 16 distinct chord combinations
  • Spectral analysis confirming theoretical models
  • Team leadership and user research integration

Technologies

Electronics
Raspberry Pi
ESP32
DSP
PCB Design