Dreamtonics Synthesizer V Studio 2 Pro: Realistic AI Vocals
- Ezra Sandzer-Bell
- 6 hours ago
- 9 min read
Japanese music software company Dreamtonics has released Synthesizer V Studio 2 Pro, a stand-alone application simulating realistic, AI-powered vocal performance controlled by MIDI piano roll notation, user generated lyrics, and models from a built-in voice database.
Announced March 21st 2025, the latest version of Synth V comes with faster rendering times, enhanced vocal realism, and new vocal production properties for fine tuning the audio outputs.
For some context, the first Synthesizer V product was released in late 2018. That means Studio 2 Pro has been more than six years in the making!
During that time, the company has been adopted by artists, producers, and songwriters around the world. In this article we'll review the latest features, with highlights from the user interface and settings panel.
I'll include some details about the new models available in their voice database, and we'll take a closer look at how professional songwriters and composers are using Synth V, with or without human singers in the loop.
We'd like to thank Dreamtonics for sharing an NFR copy of the software with our team and covering the time spent testing the product. All opinions in this article are our own.
Table of Contents
Getting started in Synthesizer V Studio 2 Pro
Synth V2 is arguably Japan's best AI voice generation DAW. If you want to dip your toe in, they currently offer a free 14-day trial of the software. This link is for your convenience. Please notice that it does not have an affiliate code attached to it.
Step 1. Download Synth V and install a few AI voice models. You'll be prompted to choose and install them during the app's onboarding wizard. However, if you skip that for some reason, you can always go back using the technique shown in the screenshot above. Click the microphone icon and then hit "manage voices".
Dreamtonics currently has 30 AI voice databases to choose from. You can listen to demos of each voice here on their YouTube channel. Each of these AI models were trained on real, sung vocals from a human singing voice, in partnership with a licensed voice provider.
You can use all of the voices commercially without additional licenses.
Step 2. Create a new project and open the settings panel. Click on audio & midi to set your MIDI input and audio output device.
In my experience, the app's settings defaulted to "none" for input and output, so I was stumped about why I couldn't get the playback or record features to work. Be sure to update these settings first, or you might run into the same problem.
Step 3. Hit the record button to record your melody with a MIDI keyboard. Alternatively, you can select "Note Creation Mode" shown in the screenshot above. Then click and drag on the piano roll to create notes manually.
I rarely use the Note Editing Mode because note creation supports the most important tasks, like changing note duration and position on the piano roll.
Step 4. Add lyrics by double clicking on the word or phoneme (word fragments). The text area will switch to an editing mode with a blinking cursor, as shown in the blue annotated region above. You can technically stuff more than one word onto a MIDI note, but you'll see this error message asking you to fix it. My takeaway was, only use one word per note when rendering but it's okay to temporarily stash a couple words in a MIDI note while I'm working out a melody, as long as I go back and clean it up.
Step 5. Navigate to the "export audio" icon shown above, assign your destination folder and update the channel from mono to stereo. Then hit the "bounce to files" button to save the track.
These are the most essential skills to practice when you're first getting started.
In the next section I'm going to go deeper and highlight some of the new and advanced features available in the latest version of Synth V. This is where you'll really get to see the full voice customization that's possible with this app.
New features available in the latest version
Synthesizer V Studio was originally available as a free, basic version and a pro version. In 2025, the classic Synthesizer V Studio Pro became their basic version and V2 Pro took the crown as the new pro version.
Both versions are available for a one-time purchase, with version 2 coming in at only ten dollars more than the basic option.
Existing users can upgrade to the latest version for $49.99. Voice upgrades range from $29.99-99.99 depending on which package you select.
Here's a summary of the standout features that I noticed in version 2.
Faster rendering speeds
Original: Standard offline rendering.
V2 Pro: Up to 300% faster offline rendering, no GPU required.
The new engine uses a technique called multithreading to generate vocals in a fraction of the time, even computers with less RAM and CPU. This removes one of the biggest bottlenecks in vocal production. This live rendering feature has historically required expensive hardware for competing products.
AI Retakes for iterating on vocal stems
Original: Retakes were limited to pitch and timbre
V2 Pro: Now includes additional vocal rhythm variations
Music producers often use randomization to go fishing for special, expressive nuance. Traditionally, an engineer working in the studio with real singers will try to record a number of different vocal takes. Later on, they would "comp" or slice together their favorite moments from each take to create the final vocal track.
Synthesizer v studio pro understood this from the beginning and always offered pitch and timbre variants, but version two now includes additional timing variations. This makes it easier to explore different rhythmic articulations and correct awkward phrasing quickly, deepening your creative flexibility in the DAW.
As shown in the screenshot above, you can access the retakes by clicking the music icon located on the far-right nav menu. Then click one of the four "AI Retakes" pads. Watch as the waveforms below your piano roll are modified instantly. Hit the playback button to listen to the differences.
Vocal Modes
Original: Timbre-only vocal mode editing.
V2 Pro: Separate vocal modes for Pitch / Timbre / Pronunciation.
In Synthesizer V, there are two separate panels with "modes" that pertain to vocals. Both control parameters that shape the voice outputs, but you'll find them in different parts of the app.
Notes Panel: Singing and rapping modes will determine whether you're controlling intonation and tone (for rapping) or vibrato (for singing).
Voice Panel: Vocal modes control timbre of the vocal tone itself. You'll find knobs for adjusting how bright, heavy, whispery, soft and hoarse a voice is.
I've highlighted them in pink and blue respectively, so you can spot the difference.
Smart Pitch Controls
Original: Manual control and pitch deviation only.
V2 Pro: Introduces Smart Pitch Controls.
The piano roll has a set of controls for changing the vocal pitch on a fixed note. Think of the green MIDI notes as the primary target pitch, while the thin white lines drawn on top of them control expressiveness like vibrato.
Smart pitch controls allow you to double click on those thin white lines to create a control node. You can drag those up and down to control the vocal pitch and override the pitch value of the green MIDI note.
They call them "smart" controls because the application automatically adjusts the vocal pitch around your chosen node position, to make sure it results in a realistic human singing voice. I've experimented with this feature at length and can confirm that it works very well.
Phoneme Editing
Original: Duration and strength sliders only.
V2 Pro: Dedicated Phoneme Timing Panel.
Instead of relying on abstract sliders, users now get a direct, visual interface for resizing and shifting the timing of consonants within each syllable. This gives precise control over articulation and rhythm.
Real world, professional use cases for Synth V
When it comes to writing music with generative AI vocal production software, people often ask "Why not record a human instead?". It's a fair question to ask and something that I'd like to address, before we get into the nitty gritty details about this app's new features.
AI singing voices in hit songs (no human vocalist)
AI singing voices have been popular in Japan for almost two decades, with virtual singers like Vocaloid's Hatsune Miku taking the country by storm in live, holographic stage performances. Her AI voice model debuted as a Vocaloid product back in 2007 and continues to be hugely popular today.
In recent years, there have been a small collection of breakout hits, like Heart on my Sleeve and BBL Drizzy, that took AI generated songs mainstream. There are also some lesser known artists, like Obscurist Vinyl, that use AI songs for comedy and entertainment purposes.
Web apps like Suno and Udio have opened mid-fidelity AI generated music to the public. Most people are using these services recreationally but some are capitalizing on them through streaming and ad revenue. This poses ethical questions about whether AI music is competing with human artists, and these AI song companies have been criticized for not licensing the music used for model training.
Crafting AI vocals with Synthesizer V instead of Suno/Udio
AI generated songs that are published and distributed without any changes are not protected by copyright. Therefore, most composers are opting for a pro audio software solution like Synthesizer V instead. DreamTonic licensed the vocal tracks that they trained their models on, making it a fair and equitable alternative to Suno and Udio.
Additionally, the vocal takes created with Synth V are copyright protected and do not require any additional royalty payments. This is true even if you publish the AI singing voices as is, without bringing a human singer in to recreate the AI tracks.
The most common use cases are still those involving a human, however. This is partly due to market conditions, as fans feel more connected to other humans. We'll expand on this in the next section and share two of the most common use cases for professional composers and songwriters using Synth V to assist in workflows where human-vocalists are still in the loop.
Sculpting vocal reference tracks for a human singer
It's more common to use a human singing voice during early and late stages of the creative process, with a tool like Synthesizer V empowering songwriters during the ideation phase. Let me explain a bit more so you understand what I mean.
Hitmakers are rarely the ones who actually sing and perform the final song. They're more likely to be composers who sell their work to labels, who manage the artist-vocalists who make the track their own.
Composers typically begin at a keyboard or guitar, humming and jotting down lyrics as they experiment with chord progressions. They might record the vocal melody themselves in an initial sketch, developing the project in their DAW over time. They can use Synth V during this initial ideation phase.
However, Synth V becomes even more valuable when it comes time to create a demo track for the target collaborator.
A male vocalist recording for a female artist, or vice versa, can use their MIDI keyboard to perform the melody and type lyrics into Synth V. Parameters for vocal performance can be carefully adjusted to create a high quality demo.
That vocal track can be sent over to a human vocalist for studio recording. Having already sculpted the vocal take with the utmost precision, the composer can communicate stylistic details they want the vocalist to retain in their performance. This cuts down on time and money wasted going back and forth with multiple takes.
Punching in to fix minor details in a human vocal take
Here's another interesting, real world use case that I've heard about from professionals in the industry. Audio engineers can use Synthesizer V to punch in and repair brief, problematic moments when their human vocalist collaborator is no longer available to record retakes in an appropriate time frame.
Vocal comping (shown above) is extremely common in music production, but it relies on engineers having enough high quality vocal stems to pull from. Autotune can be used to correct pitch issues, but what can you do when none of the vocal takes have the right expressive quality?
I spoke with a Grammy award winning producer recently who confessed that he had resorted to using AI generated voice clips on some occasions when he had no other choice. He had worked with one famous vocal artist who was getting old and struggling with slurred speech. The cost of studio time was too high and the challenge of getting a perfect take was too great.
So if vocals were unsalvageable, he was able to generate AI vocals with the same lyrics and melody. He found a vocal model that was close enough to the artist and recreated those moment, splicing them into the rest of the vocal track.
In the end, he was able to blend these AI clips seamlessly with the human vocal takes and create a coherent, publishable vocal track that the artist and label could feel proud of.
Start with a free 14-day trial courtesy of Dreamtonics
Dreamtonics has made Synthesizer V Studio 2 Pro available with a 14 day free trial so you can get a feel for it before committing. If that sounds interesting to you, head over to the free trial page here to register.
This is not an affiliate link and there are no affiliate links in this article. We hope that you found this product review helpful and insightful. As always, you can visit our blog hub to explore more free articles like this one.