Complete Guide to NEUTRINO: Free AI Vocal Synthesis in Your Browser

Before We Dive In: What Can AI Vocal Synthesis Actually Do?

If you've landed here searching for how to use NEUTRINO, chances are you want to know one thing: how do you turn lyrics and a melody into a realistic-sounding singing voice — for free? This guide walks you through everything, from the core concepts to step-by-step instructions, including a way to try it instantly in your browser with zero setup.

The short answer: NEUTRINO is a free AI vocal synthesis engine. You feed it a MusicXML file (essentially sheet music), and its AI model generates an expressive, human-like vocal performance. Unlike older voice synthesizers, NEUTRINO uses deep learning to automatically add natural-sounding breath and vibrato — no manual tweaking required.

What Is NEUTRINO? A Quick Overview

NEUTRINO is a neural network-based vocal synthesis system created by an independent Japanese developer known as Shiro. Released in 2020, it quickly became a sensation in the music production community and remains completely free to download and use.

How NEUTRINO Works: A 3-Step Pipeline

1. MusicXML Input: Load a sheet music file created in a notation editor like MuseScore
2. NEUTRINO Processing: The AI model estimates pitch (F0), volume, and spectral characteristics
3. Audio Output: An NSF-HiFiGAN vocoder renders a high-quality WAV file

In the traditional desktop version, this pipeline runs via command line. But as we'll cover later, the browser-based version lets you do all of this through a clean graphical interface.

Available Voice Models (Characters)

NEUTRINO ships with several voice models, each with its own distinct vocal character. Here are the highlights:

• MERROW: A calm, neutral female voice. Versatile and well-suited to a wide range of styles
• KIRITAN: Based on the Tohoku Kiritan voice, with a clear and airy tone
• YOKO: A soft, warm mezzo-soprano female voice
• NEUTRINO (Male): A steady, baritone-range male voice

Each model has its own sweet spot in terms of tempo and vocal range, so choosing the right one for your track can make a noticeable difference in the final result.

Using the Desktop Version of NEUTRINO: Step-by-Step

Here's how to use NEUTRINO installed on your own computer. If setup sounds like too much hassle, skip ahead to the browser-based option — it's much faster to get started.

Step 1: Download and Extract NEUTRINO

1. Download the latest version (Windows / Mac / Linux) from the official NEUTRINO site or BOOTH
2. Extract the ZIP file to a folder with no spaces or special characters in the path
3. Confirm the extracted folder contains model/, bin/, and musicXML/ subdirectories

Step 2: Create a MusicXML File

1. Open MuseScore (free) and create a new score
2. Assign lyrics to notes (select a note, then press Ctrl+L to enter lyric input mode)
3. Go to File → Export, choose MusicXML (.musicxml), and save the file
4. Copy the saved file into the score/musicxml/ folder inside your NEUTRINO directory

Step 3: Edit and Run the Batch Script

1. Open Run.bat (Windows) or Run.sh (Mac/Linux) from the NEUTRINO folder in a text editor
2. Set the BASENAME value to your MusicXML filename (without the extension)
3. Set the ModelDir value to the model you want to use (e.g., MERROW)
4. Save the file and run it by double-clicking (or executing it in Terminal)
5. Once processing is complete, your WAV file will appear in the output/ folder

Processing time varies depending on your hardware. With a CUDA-capable GPU, it typically takes under a minute. CPU-only machines may take several minutes per track.

Common Pitfall: File Paths and Encoding Issues

If your MusicXML filename or folder path contains spaces or non-ASCII characters, NEUTRINO may fail silently. Keep all filenames and paths to alphanumeric characters only — this is the single most common source of errors for new users.

Try It Right Now in Your Browser: NEUTRINO via LA Studio

If the desktop setup sounds intimidating, here's the good news: LA Studio is a fully free, browser-based DAW that lets you use NEUTRINO's AI vocal synthesis with no installation whatsoever.

How to Use NEUTRINO in LA Studio

1. Open https://la-studio.cc/editor in your browser (Chrome recommended for WebGPU acceleration)
2. From the top menu, select Add Track → NEUTRINO
3. Open the piano roll, draw in your notes, and type lyrics directly onto each note
4. Choose a voice model (such as MERROW) in the track settings
5. Hit play — the AI synthesizes the vocal in real time

Where the desktop version requires command-line work, LA Studio lets you enter lyrics right in the piano roll and hear results instantly. You can also export your finished track as MIDI or audio.

Browser Version: Pros and Cons

• ✅ Zero installation. Just open the URL and start creating
• ✅ Piano roll integration makes syncing vocals with your arrangement seamless
• ✅ Built-in MIDI editor, mixer, and effects — finish a full track without leaving the browser
• ⚠️ Best experienced in Chrome (WebGPU support). Some features may be limited in Firefox or Safari
• ⚠️ Heavy AI processing may require a paid credits plan depending on usage

5 Tips for Better-Sounding Vocals

Whether you're using the desktop or browser version, these techniques will help you get more realistic, musical results from NEUTRINO.

Tip 1: Vary Note Lengths to Sound More Human

Perfectly uniform note lengths sound robotic. Try slightly shortening or extending notes at phrase endings to introduce natural rhythmic feel — just like a real singer would phrase things.

Tip 2: Leave Space for Breath

NEUTRINO generates breath sounds automatically, but adding short rests between phrases gives the engine room to insert more convincing, natural-sounding breaths.

Tip 3: Stay Within the Model's Comfortable Range

Each voice model has a range where it sounds its best. For MERROW, that's roughly F3 to E5. Pushing outside that range can degrade audio quality noticeably. It's often better to transpose your song's key to suit the model than to force notes into uncomfortable territory.

Tip 4: Add Reverb and EQ in Post

The raw output from NEUTRINO is intentionally dry. Adding reverb and a touch of delay transforms it from a demo into something that sounds like a finished record. LA Studio includes 20+ built-in effects, so you can mix without ever leaving the browser.

Tip 5: Use Pitch Correction to Clean Up Imperfections

Even AI isn't perfect — certain notes can drift slightly off pitch. LA Studio's auto-tune feature provides a Melodyne-style interface where you can visually adjust pitch on a note-by-note basis.

Integrating NEUTRINO Into Your DAW Workflow

Here's how to bring NEUTRINO-generated vocals into your existing DAW (Ableton Live, Cubase, FL Studio, etc.).

Desktop Version: Import the WAV

1. Locate the generated WAV file in the output/ folder
2. Drag and drop it onto a new audio track in your DAW
3. Align it with your backing tracks and mix accordingly
4. Apply pitch correction, EQ, and compression as needed

Browser Version: Export and Go

LA Studio handles the full workflow from vocal synthesis to mixing, so exporting to an external DAW isn't necessary. But if you prefer to work in another environment, the per-track audio export feature lets you download each track as an individual WAV file.

NEUTRINO vs. Other Vocal Synthesis Tools

Where does NEUTRINO fit in the broader landscape of vocal synthesis? Here's a quick comparison:

• VOCALOID / Synthesizer V: Commercial-grade quality with extensive character libraries, but expensive (often $100+). Deep piano roll integration is a major plus for serious producers
• CoeFont / AI-based TTS tools: Great for quick vocal or narration work, but offer less musical control for nuanced melody shaping
• NEUTRINO (Desktop): Completely free, high quality, MusicXML support. Requires a bit of setup knowledge
• LA Studio (Browser NEUTRINO): No installation needed, piano roll built in — no MusicXML conversion required

In terms of pure value, NEUTRINO offers some of the best vocal quality available in any free tool. If you've never tried AI vocal synthesis before, it's absolutely worth experimenting with.

Frequently Asked Questions

Q: Can I use NEUTRINO for commercial projects?

A: It depends on the specific voice model. Some models, including MERROW, are restricted to non-commercial use. Always check the official NEUTRINO site and the terms of use for each individual character before using any output commercially. This applies to LA Studio usage as well.

Q: Does it work on Mac and Linux?

A: Yes — the desktop version supports Windows, Mac, and Linux. The browser version (LA Studio) is OS-agnostic: it runs on any machine — PC, Mac, or Chromebook — as long as you have a Chromium-based browser.

Q: Can NEUTRINO sing in English?

A: NEUTRINO was designed primarily for Japanese phonology, and lyrics are entered in hiragana or katakana. English phonemes can be input using an XSAMPA-based notation system, but achieving natural-sounding English pronunciation is quite difficult. For English vocal synthesis, a dedicated tool would serve you better.

Q: Do I need a GPU?

A: The desktop version works on CPU alone, but processing is significantly slower — expect several minutes for a one-minute song. LA Studio leverages WebGPU for faster processing in the browser, though it will still run (more slowly) on hardware without WebGPU support.

Q: Do I need MuseScore to use NEUTRINO?

A: For the desktop version, yes — you need a notation app capable of exporting MusicXML. But with LA Studio's browser-based NEUTRINO, you type notes and lyrics directly into the piano roll, so no notation software is needed at all.

Conclusion: NEUTRINO Is the Perfect Starting Point for AI Vocal Synthesis

NEUTRINO is a remarkable tool — completely free, yet capable of producing vocal performances that rival commercial software. The desktop version takes a bit of setup, but once configured, you can generate high-quality vocal audio offline whenever you need it.

If you want the fastest possible on-ramp, LA Studio's browser-based NEUTRINO is the way to go. Open the URL, sketch out some notes and lyrics in the piano roll, and you're hearing AI-generated vocals within minutes — no installation, no configuration. And since LA Studio also includes a mixer, effects, BPM detection, and a full set of production tools, you can take a track from first note to finished mix without ever leaving your browser.