Zunda Union Icon
Zunda Union Edamame Union

Metan Taught Me Some Tricks to Make My Voice Sound More Natural in VOICEVOX, Nanoda!

Metan Taught Me Some Tricks to Make My Voice Sound More Natural in VOICEVOX, Nanoda!

Hey everyone~! It’s Zundamon, nanoda!

The other day, I got a message like this, nanoda.

“I tried making Zundamon talk in VOICEVOX, but he sounds kind of robotic…”

…I-I get it, nanoda.

Actually, when I tried to make my own voice read a long sentence, I turned into a fast-talking monster with zero breaths, nanoda.

So today, I asked the Zunda Union’s number one theorist, Shikoku Metan, to be my teacher, nanoda!

Metan: “Oh my. Zundamon studying his own voice? What a great attitude.”

Me: “I’m going to be serious today, nanoda!”

Metan: “…If only that seriousness would last for more than three minutes.”


🎤 First, Let’s Talk About VOICEVOX, Nanoda

VOICEVOX is a voice synthesis software that reads text aloud in a character’s voice when you type it in, nanoda.

And what’s more, you can use it for free, nanoda. My voice is included, and I get to be in lots of tutorial videos and game streams 💚

If you’re new to it, I think it’s easiest to start by reading our site’s Beginner’s Guide to VOICEVOX, nanoda.

For downloads and the latest instructions, please check the official VOICEVOX website, nanoda.

Metan: “Let me tell you the most important thing first. Software gets updated. Always check the official ‘How to Use’ page for detailed instructions. What we’re talking about here is just the general approach.”

Got it, nanoda!


âœ’ī¸ First, Use Commas to Create “Breaths,” Nanoda

The first thing Metan taught me had nothing to do with the sliders, nanoda.

Metan: “Before you start adjusting, fix your script. The naturalness of the reading is almost entirely determined by the text you input.

Huh, it’s not the sliders, nanoda!?

Metan: “That’s right. Even for humans, a sentence without punctuation is hard to read, isn’t it?”

That’s true. Without commas, you don’t know where to breathe, so you try to read the whole thing in one go, nanoda.

  • ✕ I’m Zundamon nanoda today I’m eating zunda mochi nanoda
  • ○ I’m Zundamon, nanoda. Today, I’m eating zunda mochi, nanoda.

Putting in a comma “、” creates a proper pause, nanoda.

Metan: “And there’s also an item called ‘Pause Length’ that determines the length of that pause itself. You can adjust how much silence there is within a sentence.”

Me: “So, I use commas to create the pause locations, and ‘Pause Length’ to decide how long they are, nanoda?”

Metan: “Exactly. That’s the most logical order.”

I totally get it now, nanoda!


🎚 Teacher Metan’s Slider Course, Nanoda

Zundamon and Metan sitting side-by-side in front of a computer, looking at the VOICEVOX screen

Now it’s time to talk about adjustments, nanoda.

Metan explained the items listed on the official “How to Use” page one by one, nanoda.

  • Speed … How fast it reads. The higher the number, the faster it gets, nanoda.
  • Pitch … The pitch of the voice. The higher the number, the higher the pitch, nanoda.
  • Intonation … The rise and fall of the voice. Making it smaller results in a robotic reading, nanoda.
  • Volume … Just what it sounds like, the loudness of the voice, nanoda.
  • Pause Length … The length of the silence within a sentence, nanoda.

Me: “If I don’t want it to sound robotic, I should just crank the Intonation all the way up, nanoda!”

Metan: “Go on, try it.”

I tried it, nanoda.

…My voice sounded like a musical actor whose emotions just exploded, nanoda.

Metan: “If you go to extremes, it just becomes unnatural in a different way. Move it little by little, and check it by ear each time. That’s the most reliable way.”

🐌 My Favorite Way to Do It, Nanoda

This is the process Metan taught me that I really like, nanoda.

  1. First, lower the Speed just a tiny bit (that alone makes it easier to understand, nanoda).
  2. Don’t mess with the Intonation too much (move it a little, then compare it to how it was before, nanoda).
  3. If you’re putting it in a video, decide on the Volume first (it’s sad to have to redo everything later, nanoda).

Metan: “Once you find a combination you like, remember it. It’s inefficient to search for it from scratch every time.”


🔤 Accent and Intonation, Nanoda

Zundamon peeking from the side at Metan’s hand pointing to the VOICEVOX adjustment bars

This is where the real work begins, nanoda.

VOICEVOX has a system for finely adjusting the high and low points of the sound, nanoda.

  • Accent … You move the bar that appears above the text left and right to decide where the pitch drops, nanoda.
  • Intonation … You move the vertical bars up and down to change the pitch of that part, nanoda.
  • Length … The length of each character’s sound. The right side is the vowel, and the left is the consonant, nanoda.

Metan: “Place names, people’s names, and made-up words. The more a word isn’t in the dictionary, the more likely its accent will be off.

Me: “Is ‘Zundamon’ okay, nanoda?”

Metan: “You’re already read correctly in most environments, I imagine. …You’ve become quite famous.”

I’m blushing, nanoda~💚

📖 Register Words You Use Often in the Dictionary, Nanoda

It’s a real pain to fix the same word every single time, nanoda.

For times like that, if you register the word in the Reading & Accent Dictionary, it will use that reading from then on, nanoda.

Metan: “Let’s see… how many proper nouns appear in your videos?”

Me: “Zunda mochi, zunda paste, zunda shake, and zunda dango, nanoda!”

Metan: “…That’s all zunda. Just register them all at once.”

The trick is to add names that appear many times in your work to the dictionary first, nanoda.

A little effort now makes everything easier later, nanoda.

Metan really loves efficiency, nanoda…


âœ‚ī¸ Cut Long Sentences, Nanoda

The last tip she taught me was the simplest and most effective, nanoda.

Metan: “Make your sentences shorter. That’s all.”

That’s it, nanoda!?

Metan: “Long sentences will sound unnatural no matter where you break them. It’s faster to cut them into two sentences with a period than to add more commas.”

I tried it, nanoda.

  • ✕ Today I was planning to make zunda mochi and was boiling some edamame when Zunko-sama came and ate my share too nanoda
  • ○ Today I decided to make zunda mochi, nanoda. While I was boiling the edamame, Zunko-sama showed up, nanoda. …She even ate my share, nanoda.

It’s totally different, nanoda!

An easy-to-read sentence is an easy-to-listen-to sentence, nanoda.


💚 Today’s Summary, Nanoda

Teacher Metan’s lesson was super easy to understand today, too, nanoda!

  • Fix your script before touching the sliders. Use commas to create places to breathe, nanoda.
  • Adjust Speed, Pitch, Intonation, Volume, and Pause Length little by little and check by ear, nanoda.
  • If you raise the Intonation too much, it sounds like a musical, nanoda.
  • You can fine-tune the pitch with Accent, Intonation, and Length, nanoda.
  • Register words you use often in the “Reading & Accent Dictionary,” nanoda.
  • Cut long sentences with a period. This is the most effective trick, nanoda.

And I’ll say the important thing one more time, nanoda.

For detailed instructions and the latest features, please check the “How to Use” page on the official website, nanoda.

The software is always being updated, so looking there is the most reliable way, nanoda → VOICEVOX Official Website

If you’re a beginner, feel free to check out our Beginner’s Guide to VOICEVOX, too, nanoda.

You can learn about my teacher, Metan, on Shikoku Metan’s page, and about me on Zundamon’s page, nanoda.

If you have any settings that made things sound more natural, I’d love it if you’d secretly tell me on X at @zunda_union, nanoda.

I hope my voice can sound even more natural in everyone’s creations, nanoda 💚

This was Zundamon, nanoda~! 🎤