Act 4: Final stretch

So you've given your character a face and a personality. But when they speak, they still sound like everyone else. That's because you haven't given them a voice yet.

Welcome to the final stretch. In this act, we're covering the audio settings that make your character actually speak, and the final touches that determine whether they stay in your private collection or go out into the world.

---

1) Custom Voice Upload

Scroll down in Stage 4, and you'll find the Audio Section. By default, this is collapsed. Click the header to expand it.

This feature lets you give your character a custom TTS voice by uploading an audio sample. The system uses that sample to clone the voice, which is then used whenever the character speaks in chat.

The Upload Process

You have two options:

• Upload Button: Click the small upload button to open a file picker. Accept standard audio formats (MP3, WAV, etc.).
• Drag & Drop: If you're on desktop, just drag an audio file directly onto the designated area.

Once you upload a file, you'll see a waveform visualization appear. This is where the actual work happens.

---

2) Audio Requirements and Trimming

You can't just upload a 3-minute rambling clip and call it a day. The system requires a specific clip length: 10 to 30 seconds.

Why? Shorter clips don't have enough data for the model to learn the voice. Longer clips introduce noise, inconsistencies, and processing issues. The sweet spot is a clean 15-20 seconds of clear speech.

What Makes a Good Sample

• Clear speech: No background music, no heavy static, no overlapping voices.
• Single speaker: Don't use clips with multiple people talking.
• Natural tone: Avoid exaggerated acting or whispering. The model replicates what it hears.
• No long pauses: Dead air wastes your precious seconds.

Trim your clip down to the best 10-30 seconds, then move on.

---

3) Voice Sample Transcriptions

Below the waveform, there's a text input labeled Audio Sample Text.

This field matters more than you think.

Why Transcription Helps

The TTS system uses your transcription to align the audio with the text. A good transcription:

• Improves pronunciation accuracy
• Helps the model understand pacing and emphasis
• Reduces artifacts and glitches in the final output

How to Do It Right

Listen to your trimmed clip. Type out exactly what is being said, word for word. Don't paraphrase. Don't summarize. If the speaker says "um" or stutters, include it if you want that natural feel, or trim those parts out of the audio instead.

A sloppy transcription leads to a sloppy voice. Take the extra minute to do it correctly.

---

4) Plan Restrictions

This feature isn't available to everyone.

If you see a lock overlay on the audio section, it means your current plan doesn't support custom voice cloning. An upgrade button will be present if you want to unlock it. Free users and lower-tier subscribers won't have access to this tool. The reasoning here being that it's a tool that is scrutinous. A payment is essentially a proof of age.

---

5) Stage 3: Final Touches

You've made it to the final stage. This is where you organize, test, and decide your character's fate.

The interface here is straightforward, but each element serves a specific purpose.

---

6) Tagging Your Character

Tags are how people find your character. They're also how the system categorizes content.

The Tag Input

Type a tag into the input field and hit Enter or click the plus button. Tags appear as pills below the input. To remove one, click the X on the pill.

Popular Tags

Below your tags, you'll see a section labeled Popular Tags. These are quick-add buttons showing commonly used tags and their usage count. Click one to add it instantly.

Best Practice

Use tags that actually describe your character. "Fantasy," "Romance," "Sci-Fi," "Villain," "Assistant". These help users find what they're looking for. Don't tag a vampire accountant as "Action" just for visibility. You'll annoy people, and annoyed users don't come back.

---

8) Public vs. Private Visibility

At the bottom of the final stage, you'll find the visibility toggle.

The Toggle

Flip it to make your character Public. This adds them to the searchable database for all users. Keep it off, and the character stays Private. visible only to you.

Restrictions

Not everyone can make unlimited private characters. It costs a lot to store encrypted characters, so we impose some limits on free users.

• Free Users: You are limited to 3 private characters. If you try to create a fourth, you'll need to make an existing one public or delete it.
• Cloned Characters: If you imported a character (using the CharacterHub import or cloning another user's bot), you cannot make it public. This is to prevent spam and attribution issues. The toggle will be disabled with a notice explaining the restriction.

If you hit your private limit, an info message appears. Mobile users will see an upgrade prompt if they want more private slots.

---

9) Age Detection and Safety

MiocAI has automated systems to detect characters that violate our safety policies. This is non-negotiable.

What Gets Flagged

The system scans character data for indicators of underage content.

What Happens When Detected

You wont instantly get banned or such, it could have been a system error after all. We do not log these errors. If the system flags your character:

1. A modal appears: "Child-like Character Detected"
2. Creation is blocked until the issue is resolved.

There are no workarounds, and there is no appeal process for this specific restriction. If your character is flagged, you need to revise the description, backstory, or age.

We covered this in Act 1, and it bears repeating: Do not create underage or childlike characters. Not for SFW contexts, not for "wholesome" stories, not for any reason. The variables are too complex, the risks are too high, and the answer is always no.

---

10) TLDR

Do

• Trim your audio clips to 10-30 seconds of clean, single-speaker audio.
• Transcribe accurately to improve voice quality.
• Use relevant tags that describe your character honestly.
• Test example messages to verify tone and personality alignment.
• Respect the age detection system. It exists for a reason.

Don't

• Upload messy audio with background noise or multiple speakers.
• Skip the transcription field.
• Misuse tags for visibility.
• Try to bypass safety filters. You will lose.

---

You now have the complete picture. From the first click of "Create" to the final toggle of the visibility switch, you know how to build a character that looks right, acts right, and sounds right.

The tools are in your hands. Use them well. And don't make anything that makes us regret giving them to you.