PngTuber Maker is a Steam application by YiMeta that makes an avatar react to sound using PNG images. To set it up, install the correct app from Steam, add at least two expressions, choose a microphone and adjust the sound threshold while testing your voice.
The name can be confusing: this guide is about the Steam app called PngTuber Maker, not a generic browser avatar tool or another similarly named programme. Its virtual-camera output may suit a streaming setup, but check that your broadcast software can capture it before you plan a live session around it.
Confirm you have the right PngTuber Maker
Search Steam for PngTuber Maker and check that the publisher is YiMeta. The product page is the reference point for identifying this particular app: PngTuber Maker on Steam. Streamlabs also has a tutorial titled How to Use PngTuber Maker from Live 3D, which points readers to the Steam download. The similarly worded “Live3D” in a tutorial title does not make every product using those words the same application.
This distinction matters because avatar tools can share familiar terms—PNG, reactive, virtual camera—while having different installation steps and output methods. If the app title or publisher does not match, pause before following instructions for it. A tutorial for a browser tool or a separate avatar application may not describe the controls you see in YiMeta’s Steam app.
The basic workflow is straightforward: provide the artwork, designate the microphone that should drive sound detection, then set the threshold so the avatar changes when you speak rather than whenever the room makes noise. The threshold is not a universal magic value. A laptop microphone in a quiet room and a USB microphone beside a fan will not behave identically.
Check the current Steam listing before installing
Store pages and software requirements can change. Open the listing at the time you intend to install and verify that the app remains available, that the listed operating system and hardware suit your computer, and that the current installation path still matches your expectation. Do not treat old screenshots, a search result snippet or a tutorial recorded earlier as proof of current compatibility.
The listing has described PngTuber Maker as free software and has included Windows requirements. Those are store-page statements, not a guarantee that every current Windows device or configuration will work. Check the live Steam listing and system requirements directly. If you use a managed work computer, check whether you can install Steam and run the app under your account before planning a broadcast.
Once installed, launch the application and take time to recognise its current interface before adding it to a show. Avoid basing a setup on a remembered menu label: interface layouts and device names may differ across updates or computers. If a tutorial’s controls do not match what you see, refer back to the current listing and app guidance rather than assuming the missing option is hidden in a particular place.
It is useful to separate two checks. First, can the application launch and show your avatar locally? Second, can the broadcast software receive the avatar output in the way you need? A successful first check does not establish the second. The Steam listing describes a virtual-camera feature, but the evidence does not establish that every current streaming programme captures it identically.
Prepare PNGs with distinct expressions
Start with at least two PNG images of the same avatar in different expressions. The core idea is a neutral or resting image and a speaking image. Keep the character’s position, scale and framing consistent between them; if the face jumps across the canvas whenever the state changes, it can look like a glitch rather than a deliberate response.
Use expressions that remain legible at the size viewers will see. A slight mouth opening may disappear when the avatar is reduced to a corner overlay. A clearer open-mouth pose or a change to the eyes can make the speaking state easier to read. Do not add detail simply to make the files more elaborate: clean outlines and a consistent crop are more useful than tiny features that vanish on a phone screen.
Before importing, check that the files really are PNG images and that the transparent areas are intended. A visible white rectangle behind one pose and a transparent background behind another can make switching look uneven. If you want a solid backdrop, it is often simpler to use the same background treatment for each state or add a background in the app if that option is available in its current version.
The Steam listing describes adding at least two PNG expressions; additional state images can make the effect more varied. Begin with the minimum pair, because it is easier to troubleshoot one transition than a large set of states. Once speech detection works, add expressions only if each one has a clear purpose, such as a closed-mouth resting look and several speaking or reaction poses.
A useful preparation check is to open each image in an ordinary image viewer before importing. Confirm the face is not cropped, that the transparent canvas is not unexpectedly huge, and that the expressions differ enough to notice. Save copies of your originals in a separate folder. That gives you a clean fallback if an edit makes the imported version hard to distinguish or correctly frame.
Select a microphone for sound detection
Choose the microphone you actually expect to use while speaking. A computer may expose an internal mic, a webcam microphone, a headset and a USB microphone at once. Selecting the wrong one can make the avatar stay still even though your voice is audible to you through another device or application.
Connect the intended microphone before configuring the app, then select it from the available input devices. If the device has more than one similarly named entry, speak briefly and watch whether the input responds before proceeding. You do not need a special model just to try the setup; a stable, intelligible voice input is more important than buying equipment before you know what your channel needs.
Position the mic consistently. If it is close to your mouth during testing but farther away during the broadcast, the amount of sound reaching the app changes and the threshold may no longer behave as expected. Likewise, avoid pointing a sensitive mic at a fan, keyboard or loud speaker when practical. The avatar responds to detected sound, not to the meaning of your words, so unrelated room noise can trigger it.
For a channel that is live continuously, consider whether the avatar is meant to react only during spoken segments or also during music and other audio. If the app listens to a microphone while your channel’s programme audio plays through speakers nearby, the mic may pick up that playback. Headphones or separating the narration mic from the programme speakers can make the test more predictable. This is a workflow choice, not a reason to assume a particular device setting will solve every case.
Adjust the sound threshold gradually
The threshold determines how much detected sound is needed before the avatar changes state. If it is too low for your room, background noise can keep the speaking expression active. If it is too high, ordinary speech may not register and the avatar can remain in its resting pose. The appropriate setting depends on mic level, distance, gain and surrounding sound; no single numeric value is reliable for every setup.
Start with a quiet room and speak at the volume you expect to use on stream. Adjust the threshold in small increments, testing after each change. Aim for a setting where ordinary speech triggers the speaking expression, while a short pause returns it to the resting expression. If the interface provides a visible input indication, use it as a clue rather than as a substitute for observing the avatar itself.
If the avatar reacts constantly, first identify the source before raising the threshold repeatedly. Check for a fan, music from speakers, keyboard noise or a second microphone receiving sound. If speech fails to trigger it, verify the selected device and its input level before lowering the threshold. This order helps avoid masking a device-selection mistake with a sensitivity adjustment.
A short spoken test is better than shouting into the mic. Test at the usual distance and volume, pause, then repeat a sentence. A threshold that only responds to a loud close-up test is not suitable if you speak softly at a normal distance. Conversely, a threshold that activates from every chair movement will be distracting during quiet moments.
Test the avatar response before streaming
Test one variable at a time: first confirm the resting image, then speak and check for the speaking image, then pause to see whether it returns. Repeat at your real speaking distance. If the avatar fails to switch, check the selected microphone and threshold; if it switches but looks wrong, inspect the PNG crop, transparency and expression contrast.
Next, test the exact capture route you intend to use. The Steam listing describes virtual-camera output. Open your broadcast software and see whether it can select that output, then confirm the avatar is visible in the preview. Capture names, permissions and device handling differ between programmes, so do not assume that a setup described for one workflow applies to yours. Streamlabs’ tutorial can provide a starting point, but confirm the current controls in the software you actually use.
Check the result at the size and placement planned for the broadcast. A large preview can hide a problem that is obvious when the avatar sits beside a chat box or over a video. Verify that expressions do not obscure important information and that any background or accessories remain consistent. PngTuber Maker’s listing also describes backgrounds, avatar accessories, switching animation and desktop mascot use; treat these as optional customisation, not required steps for basic sound response.
If the avatar is part of a prerecorded loop or an always-on channel, decide whether it belongs in the programme at all times or only during a host segment. The app’s microphone-driven response is designed around detected sound; it does not turn a prerecorded video into a live presenter. For a continuous video programme, first get the underlying playback workflow reliable. Our guide to keeping a prerecorded YouTube stream running from a VPS or spare PC explains the separate choice of where that loop runs.
A desktop test is not the same as a live YouTube check. Before a real broadcast, use the capture method you have verified and make a private or otherwise suitable test if your channel workflow allows it. Look at the viewer-facing output, not only the app preview: framing, audio routing and the captured camera can differ. For a video-loop channel, the FFmpeg loop-stream command guide covers a different task—repeating a video source—so keep that separate from avatar sound detection.
Decide whether this belongs in your channel workflow
PngTuber Maker may suit a channel where a person speaks on camera-adjacent segments but prefers an illustrated avatar. It can also support a desktop mascot use, according to the listing. It is less relevant if your channel is only a continuous devotional or ambience recording with no live microphone narration. Choose the tool for the job you actually have, not just because the avatar changes when sound is detected.
For a broadcast that depends on a computer staying on, the avatar app is one part of a wider chain: the app must run, the correct mic must be available, the capture path must be present, and the stream programme must continue. If you need a separate overview of operating choices for a non-stop video feed, see how a YouTube 24/7 streaming service works with a playlist. If your channel is a recorded worship programme, planning a worship set without silence between videos addresses continuity between source clips rather than avatar setup.
If a PC must remain available simply to keep a video broadcast running, StreamNeo removes that specific burden for an uploaded video channel: you upload the video once, provide your YouTube stream key, and the broadcast can continue with your computer switched off. It is YouTube-only, so it is not a way to host or run PngTuber Maker itself; keep the avatar workflow and the video-loop workflow distinct.
Before committing, compare the operating options on the pricing page. When the file and channel are ready, start free — 24-hour trial, no card.
FAQ
Is PngTuber Maker the same as Live3D or Streamlabs Live3D?
This guide concerns the Steam app PngTuber Maker by YiMeta. The Streamlabs tutorial uses the phrase “from Live 3D,” but that does not establish that all similarly named tools are the same product. Check the app publisher and current listing before installing.
How many PNG expressions do I need to begin?
The listing describes a basic setup using at least two PNG images with different expressions. Start with a resting and speaking pose, then add more only if you want additional states and can distinguish them clearly at stream size.
What threshold should I use?
There is no one setting that suits every microphone and room. Adjust it gradually while speaking at your usual distance, and check both that normal speech triggers the avatar and that quiet pauses return it to the resting image.
Will the virtual camera work in my streaming software?
The listing describes virtual-camera output, but that does not confirm compatibility or identical behaviour in every broadcast programme. Select the output in your own software and test the viewer-facing preview before relying on it for a live session.