Lucent's Guide to Online Sequencer - Vocal Synthesis Using Percussion

Welcome to Lucent Guide v1.2! This site was last updated on September 21, 2022. Site is currently on hiatus indefinitely.

Lucent's Guide to Online Sequencer

Vocal Synthesis Using Percussion

Preface

Using the same concept as frequency chaining, one can generate vocals on OS. However, they are difficult to recreate by hand and require an external tool to easily automate frequencies for sequencer use.

Vocals are made up of a range of frequencies (as opposed to being isolated to one tone)—on top of constantly shifting pitch to ever pinpoint accurately. If you’re interested in this jargon, I suggest reading up about it here.

Chances are, you have seen an MP3 to MIDI conversion sequence on the site. This kind of conversion analysis is known as FFT—in which notes are assigned across the piano roll in order to best represent the audio sample. Audio clips generated this way tend to be very messy and rather impractical to use in a legitimate song.

Instead, we will use a PCM converter to generate audio samples on OS. I asked the user Frank how he was able to automate the drum frequencies for this sequence. In response, he released a user-friendly version of the PCM converter he programmed for it, which I’ll explain how to apply in this article.

Through PCM, audio is also generated physically from grouping tiny notes at a specific rate. It is more accurate than FFT conversion as it is able to represent all frequencies outside the 12-tone equal temperament system.

The other benefits to PCM is that the clarity is great and doesn't require cleanup unlike FFT converted samples. PCM samples are also flexible across multiple drum kits allowing for various sound textures—whereas FFT isn't, being limited to 8-bit Sine.

The downside to using PCM is that it outputs a mass number of notes in order to replicate note frequencies. OS has a tendency to lag and break audio under high-density situations, therefore making this one of the more difficult techniques to control. After conversion, this technique requires a lot of post-maintenance in order for your sequence to play properly.

This article will guide you on how to obtain PCM samples, as well as handing out some general tips and discussing other miscellaneous items that are related. This technique requires a strong computer to run—if you have difficulty loading some sequences, it is recommended you do not use it.

For reference, here is a song that I've made using percussion-chained vocals. It showcases how to effectively use these vocals on OS, as well as gauging how intensively it will run on your computer.

Preparing the Audio File

To start using Frank's PCM converter, first we need to prepare a compatible audio file.

Obtain Audio File

Use only vocal samples with no background music. PCM conversion attempts to replicate singular tones and does not work well with converting multiple voices.

(I used a Vocaloid sample here, which generates ideal vocals for this technique)

Download Audacity

Audacity is an open-source audio tool and you’ll need it for the following steps.

Preparing Samples

If you have a long recording, I recommend splitting the sample into multiple sections. Allowing silence between phrases makes them easier to access and edit when you import them on OS.

You can highlight a section and hit Ctrl+X to cut it from the recording. Then, click anywhere on the canvas to assign the playhead (moving vertical line) at a new location—before hitting Ctrl+V to paste.

Allow at least one second of silence between each segment. You can use the Time Shift tool to push sections around.

After separating each clip, you can highlight any unwanted parts or excess silence and press Ctrl+K to trim them from the sample.

Additional Audio Editing (Optional)

This step is for applying any of Audacity’s audio functions if you want to achieve a certain effect with the sample.

Convert from Stereo to Mono

First, make sure the track is selected.

Then, navigate to Tracks > Mix > Mix Stereo down to Mono and mix them down. You should see the two layers of the track converge to one layer.

This step is crucial if you want the converted file to work correctly. If you don't do this step, the conversion will play at half speed on OS which is unintentional.

Reduce Gain

This is to fully optimize the audio file in order to reduce output static on the conversion.

Tracks should be moderately quiet before exporting. I suggest lowering the gain to -5 dB as a baseline and depending on the scenario—make further adjustments in intervals of 2 dB.

If the sample is too loud, the OS result will sound like forced static and will glitch out more easily.

However, if the sample is too quiet, the PCM converter won't pick it up very well and some sections will not appear on OS.

Makeup Gain (Optional)

This step is for samples with segments that are too quiet in relation to louder parts (these may be cut off in the finished product).

You can highlight quiet parts and go to Effects > Amplify to close the volume gap. Work in +2 dB increments to avoid making drastic volume changes.

Export as 32-bit .wav

The converter tells you to export to 32-bit. If you don't change the file extension, there will be an error message.

Checklist

Track has been mixed down to Mono settings, not Stereo
Master gain has been reduced by at least -5 dB
File has .wav extension and is 32-bit, not 16-bit

Once those settings have been checked, go ahead and upload the file...

...and you should get a file like this, just drag it over the sequencer.

Import On OS

By default, the file will be a 300 BPM chain of Drum Kit High-Q notes. As a suggestion, you can turn them into Electric Square Clicks which I find cleaner.

For now, you're only able to compose PCM songs with 150 or 300 BPM in mind. Like frequency chaining, changing the BPM will cause notes to become out of tune. Fortunately, you can calculate the speed changes to apply on the audio file before importing it on OS. I will mention this in a later section.

Post-processing

Now that we have our audio sample, you can either save it and call it a day—or if you’re like me and you want to make something legit, it’s imperative that we go through the hell on earth process of optimizing it so that your computer can play something audible.

A forced audio break occurs when your soundcard is unable to handle the density of notes that are being played at once.

To avoid causing forced audio breaks on your computer, you’ll need to decrease the overall density of the samples. You can do this three ways.

Even Distribution

Start off evenly removing notes by holding right click and running over the sample. Don't worry about doing this incorrectly—the sample is extremely dense and you won't hear audible changes for the first few runs.

I would repeat this step up until you can see new note names appear. If you can visibly see the labels, you may have removed too many notes.

Fair amount of cleaning.

Too much!

What your computer can process will vary from person to person. Remember that removing more notes will make your sample less accurate and become more stuttery.

Fading Tails

Next, you can scrub the tails of the sample to fade them out. It is okay to sacrifice audio quality for the ends of notes as they will fade with the background music. Ideally, you'd want the initial sound of a note to be the clearest.

This is particularly useful for prolonged vocal notes. The longer the PCM chain is, the more audio breaking buildup that will be incurred.

Vocal Chops

You can manipulate segments of a sample on OS to create vocal chops. Copy, clean, and reproduce vocal chains however you wish.

Because vocal chops only use part of a sample at a time, they’re also very good for density reduction. Chopped chains tend to do better in performance.

Other Tips Using PCM Samples

If you find the vocal sample to be too quiet in relation to the other instruments, you can also use the console to amplify the volume past max value.

Drum Kit at 400% Volume (OS max default is 200%)

setInitialInstrumentVolume(2, 4)

Electric Drum Kit at 400% Volume

setInitialInstrumentVolume(31, 4)

You can switch between drum kits to alter the texture of the voice. This is somewhat useful for detailing if you have extra kits to spare.

Additional Audio Editing

This segment continues Step 4 of editing in Audacity. I will mention some important features using the program to customize your sample before import.

Adjusting Sample Speed

Frank's PCM converter will only export to 300 BPM. However, you can change the speed of the original file on Audacity using the Change Speed tool.

Locate “Change Speed” under the Effect tab of Audacity—make sure it’s the one that changes both tempo and pitch.

To obtain the value of the speed multiplier, use:

300/Final BPM

If I wanted a final BPM of 250, then my speed multiplier would be x1.2.

I recommend working at an enlarged BPM (250 instead of 125) as working with the smaller BPM will make the sample twice as dense.

You can export and convert the file like normal. Once you import the modified file on OS, you can change the sequence BPM as desired.

I've made the example speed change in the first sequence. You can compare it with the original 300 BPM sequence on the right. Note that the audio sample is made with 300 BPM in mind and that the notes do not align with the 250 BPM grid.

Rising Vocal Chops

It is possible to create “rising vocal chops” using the Sliding Stretch function of Audacity. This is an exponential change speed function under Effect.

First, make duplicate chops on the canvas before applying the function. Then, separate each chop like below. This will make them easier to edit in OS.

I've made an example sequence here. Exporting the file will not be perfect and that you will need to do more post-OS editing in order to align the chops to the grid, as well as other PCM housekeeping items.

Page updated

Google Sites

Report abuse

Vocal Synthesis Using Percussion

Preface

Preparing the Audio File

Obtain Audio File

Download Audacity

Preparing Samples

Additional Audio Editing (Optional)

Convert from Stereo to Mono

Reduce Gain

Makeup Gain (Optional)

Export as 32-bit .wav