Tag: mastering

Basics of Mixing – 14.3 Oversampling and Upsampling

Hi! This is Jooyoung Kim, a mixing engineer and music producer.

During my undergraduate studies in physics, I often used my extra credits to take music courses. Looking back, I regret not taking any Python classes—especially now that I’m studying plugin development, data processing, and methodology. Without AI tools, I wouldn’t have been able to start coding at all.

Currently, I’m in the final semester of my master’s program in the Department of New Media Music. I’m not too worried about my thesis, so this semester, I’m focusing on a personal project: developing a saturation plugin in my own way. I know it will be challenging, but I also aim to write a paper introducing a new methodology for building audio plugins.

Now, let’s talk about oversampling and upsampling. This article is based on my book, ‘Basics of Mixing‘, released in South Korea.

What are oversampling and upsampling?

Many audio plugins offer oversampling, but what exactly is it, and how does it work?

  • Oversampling: Increasing the existing sample rate by an integer multiple (e.g., ×2, ×4, ×8, etc.).
  • Upsampling: Increasing the existing sample rate, but not necessarily by an integer multiple.

In oversampling, the process inserts zeros into the empty values and interpolates them using a low-pass FIR (Finite Impulse Response) filter..

It’s me! The right one is the photo on the left doubled horizontally.

Think of it like stretching an image by an integer factor: the blank spaces are filled in through interpolation. After processing the audio, the plugin then downsamples the result back to the original rate.

On the other hand, upsampling converts the sample rate to another rate through interpolation. This is a different process from oversampling.

Why do we use oversampling and upsampling?

Oversampling helps reduce aliasing and creates a more natural-sounding result, while upsampling is primarily used for changing the sample rate.

However, does an oversampling function always make a plugin sound better? That’s a different discussion.

For example, the SSL Bus Compressor 2 plugin provides an oversampling option. Below, you can see the frequency response of the plugin:

The left graph shows the response without oversampling. The right graph shows the response with 2× oversampling.

Both graphs exhibit harmonic distortion at the same positions. While other measurements appear similar, even small differences between the two can result in noticeable changes in sound. In my experience, non-oversampled processing often sounds better.

That was a brief explanation of oversampling and upsampling! See you in the next post!

Basics of Mixing – 14.2 Aliasing and Nyquist-Shannon Sampling Theorem

Hi! This is Jooyoung Kim, mixing engineer and music producer.

The paper I submitted to JASA is currently under review.

The assigned editor is a renowned scholar in the field of acoustic signal processing at Tsinghua University. I’m more nervous than I was during my college exams..

By the way, today we will talk about the aliasing and sampling theorem – the basic knowledge for mastering audio.

Let’s start!!

Aliasing

Aliasing is a phenomenon caused by converting analog data to digital data. You can see the same picture; the left one looks normal, but the right one appears distorted. The right image was made by resizing left image in low resolution.

Why does it happen?

Nyquist-Shannon Sampling Theorem

A study on factors affecting transmission published by Nyquist in 1924. Reading the paper, it seems that there were studies at the time that showed that waveforms such as sine waves, triangle waves, and square waves had a significant impact on transmission.

But the courageous Mr. Nyquist says in a strong tone that this is not the case…chill guy..

Nyquist laid the foundation for sampling theory through his research on telegraphy in 1924 and 1928.

Claude E. Shannon expanded on this in his 1949 paper, ‘Communication in the Presence of Noise,’ which solidified the modern sampling theory. If you are curious about what kind of paper was actually written, I have included a link so please take a look..

For those working in music and audio, the finer details aren’t crucial, so I’ll focus on the key points.

1) Only frequencies up to half the sampling rate can be accurately represented (this is known as the Nyquist frequency).

2) The original source with a value higher than the Nyquist frequency is expressed as a value below the Nyquist frequency when sampled (this is the aliasing phenomenon).

3) Therefore, if you filter it in the high-pass to filter out this noise, it becomes clean (this is called the anti-aliasing filter, The original Nyquist paper made telegraph transmission easier through this filtering).

The blocky artifact shown in the first image is also a form of aliasing caused by sampling errors. Since it samples space, it is a bit different from the sound of sampling time.

An ADC (Analog-to-Digital Converter) chip performs these tasks!

High-end ADCs and DACs improve the precision of this process.. Still, Lavry Gold converters are extremely expensive… Maybe due to low demand?

The first part of this video shows how aliasing sounds when you lower the sample rate. Using a filter before this plugin to remove frequencies near the Nyquist limit will significantly reduce aliasing noise.

That’s all for today.

In the next article, I will talk about oversampling and upsampling.

See you in the next post!

Basics of Mixing – 14.1 Pre-Mastering Audio

Hi! This is Jooyoung Kim, mixing engineer & music producer.

Before, we talked about hardware. Today, I’ll begin by introducing mastering. This article is based on my book, Basics of Mixing, published in South Korea.

Let’s dive in!

What is Mastering?

Why do we master audio?

  1. The loudness of mixed music is too low
  2. So gain more volume for the mixed music, maintaining the instruments’ balance

Of course, mastering involves many aspects. You can easily understand mastering with the two sentences above.

Simply put, mastering is the final step in music production. It can also be described as packaging the mixed music. The package does not change after mastering.

What a simple and beautiful setup! – https://gearspace.com/board/mastering-forum/1252793-mastering-studio-photo-thread.html

Room acoustics are the most important factor in building a mastering studio. Also, great quality ADC(Analog-Digital Converter)/DAC(Digital-Analog Converter), speakers and hardware are necessary too.

Building a mixing studio is also expensive. However, in mixing studio, audio hardwares take quite large volumes-they make room acoustics worse.

A desk with hardware may look good and fancy, but it takes up a lot of space. This can lead to comb-filtering effects. Removing the desk is a great solution, but as you know, it’s easier said than done..

One more thing, a mixing engineer listens to a song too many times, making it hard for them to recognize problems. Therefore, when you need to master a song or an album, I recommend hiring a mastering engineer.

Nevertheless a mixing engineer should provide a louder monitor file to clients than the mixed file. Due to budget constraints, mixing engineers sometimes have to master their own files.

In this chapter, for those situations, I’ll talk about the process and concepts about mastering.

The full details start next post! 🙂

2024 Year in Review

Hello! I’m Jooyoung Kim, an engineer and music producer.

It’s already the last day of 2024…
It feels like I’ve barely done anything, yet the end of the year has come around again. 😊
This year, I want to take a moment to look back on what I’ve done.

First Half of the Year

In January, I started writing introductory posts about mixing techniques.
I’d been wanting to publish articles elsewhere, but after receiving rejections from various outlets, I decided to focus on writing consistently on my blog instead.

I lost count of how many times I revised those posts! Eventually, I submitted them to Kyobo Bookstore’s POD (Print on Demand) service. To my surprise, Joongpil Goo, an engineer from Klang Studio with extensive experience as an SM engineer, graciously wrote a recommendation for the book. Even now, it feels surreal! 😊

This was finally completed by late September, and it was quite an intense journey just to publish a single book.

From January to April, I worked with a studio called Studio Dolphin, producing YouTube videos where AI collaborated with indie musicians to create songs.
We used Soundraw for the AI-generated music, ChatGPT for lyrics, and melodies were composed by indie musicians. The challenge was to create a full song in just 90 minutes.

I participated as both an engineer and a panelist, but unfortunately, the videos didn’t garner much attention, and they’ve since been taken down.

At the start of January, I was extremely busy with various recording and mixing tasks for school projects. In February, I spent a lot of time preparing paperwork to apply for government-sponsored projects. Sadly, none of these applications were successful, which left me feeling rather deflated.

In March, I returned to school, and things got busier from there.

In May, I launched my English blog and started promoting Plugin Boutique products more actively, which kept me even busier.

Summer

From July to August, I worked on producing Danny Boy’s EP from start to finish.
I handled everything—recording, arranging, mixing, and mastering. 😊

At the same time, I collected measurement data for my thesis. The data turned out well enough that I’m now preparing to submit it to an international academic journal.

While processing the data, I found myself delving into coding—a field I never thought I’d explore! Surprisingly, tools like Copilot and ChatGPT were lifesavers. At first, I barely understood anything, but after asking countless questions, things started making sense. 😊

Second Half of the Year

September was tough. I applied for several positions but didn’t get any of them. My final interview was with the Gyeonggi Arts Center, but, well… things didn’t work out. Back in May, I’d also made it to the final round for a position at Kumho Art Hall at Yonsei University but was ultimately rejected. I guess that’s just how life goes. 😊

In October, I filmed a music video, and November passed without anything particularly noteworthy.

In December, I wrote a short paper about loudness in audio and submitted it to a domestic academic journal. It was accepted and will be published in early January.

Also, during this semester, a professor from Sorigeo joined the faculty, and I took his class. In collaboration with him, I gathered data for another research paper, spending two days on measurements. The data looks promising, but I haven’t organized it yet. Once I do, I plan to publish another paper and share the details. 😊

In addition to all this, I continued working on my own music, doing external mixing/mastering, composing/arranging, and providing guide vocals.

Unforgettable Projects

One of the most memorable projects this year was mastering the EP for the Bulkuksa Team’s Random Game. My favorite track was definitely “Microdust.”

A heads-up: the song contains explicit language. 😊

A friend had asked if I could master the track using only an MP3 demo, as they’d lost the original files. I was initially concerned about the sound quality but was blown away by how creative the music was. I couldn’t say no!

Some tracks were well-balanced, while others were completely off, but I managed to improve them significantly through proactive mastering. While the sound quality wasn’t ideal due to the limitations of MP3, the sheer fun of the music made up for it. 😊

Gear Purchases

  1. Stam Audio SA-4000 MK2 Compressor
  2. KEF Kube 15 MIE Subwoofer
  3. Stam Audio SA-2A Compressor
  4. Earthworks M30 Measurement Microphone
  5. Rack Cabinet and Patch Bay
  6. Wharfedale D310 Passive Speakers
  7. Cort A4 Bass
  8. A broken Heritage Audio Successor Compressor

I bought quite a bit this year…

Next year, I’m thinking of adding a tube microphone preamp and an EQ. Maybe another compressor or a mic too? The gear wishlist never ends! 😊