Synthesizer V, ACE Studio or Vocaloid: which AI singing software fits your song?

You open a vocal synth for the first time and the screen looks like a sequencer: a piano roll, a voice picker, a lane under each note waiting for a word. You draw eight notes, the editor fills them with a placeholder syllable, and you press play. A trained voice sings a melody you wrote, perfectly in tune, on a syllable that means nothing. That is the moment most people realise what this software is. It sings exactly what you give it. It will not write the tune for you, and it will not write a single word.

Synthesizer V, ACE Studio and VOCALOID all work that way at their core. They are instruments, not songwriters. You supply a melody and a lyric, note by note, and they turn it into a vocal you can shape down to the vowel. That control is the reason to buy one, and it is also the reason to pick carefully, because the three differ in price model, languages, where the rendering happens and what else comes in the box.

Which reader this page is for

There are two very different people searching for "AI singing" tools. One wants to type a lyric, pick a style and get a finished song back with a voice already on it. That reader should start with our page on which AI singer can sing your own lyrics, which covers full song generators and how to choose a voice there. The other reader wants to decide every note: a songwriter making a demo to send to a real singer, a producer who needs a topline sitting exactly on the grid, a composer who wants a choir pad that pronounces actual words. This page is for the second reader.

If you are a producer weighing whether a synthesized vocal can stand in for a hired one on a release, the trade-offs (timing, diction, rights) get their own treatment in can an AI vocalist replace a session singer. Here the question is narrower: of the three editors people search for by name, which one fits your song.

One more thing up front, so it does not sit hidden at the bottom. sunomarket.com is an independent site. We are not Dreamtonics, not ACE Studio's developer, not Yamaha, and not Suno, and we are not affiliated with any of them. Our lyrics generator does one job in this workflow, the words, and it comes up later where it actually fits.

The three side by side

Everything in this table was checked on each company's own pages at the time of writing. Prices change, editions get renamed, and one of these products has a new major version due within weeks, so treat the numbers as a snapshot and confirm on the vendor's site before paying.

Read the table for what it says about working style, not just cost. Synthesizer V and VOCALOID behave like classic software: you pay, you install, it runs on your machine. ACE Studio behaves like a service: it renders in the cloud, it bills on a plan, and it keeps adding tools around the voice.

Synthesizer V Studio 2 Pro: buy it once, keep the rendering at home

Dreamtonics sells Synthesizer V Studio 2 Pro for $99 as a one-time purchase, and that price includes one voice of your choice. Further Dreamtonics voices are $79 each, and a bundle of the editor plus one extra voice is listed at $149. The FAQ is explicit that there is no subscription: updates within the same major version are free, and nothing is charged after the trial ends.

The free trial needs no card and covers nearly all features, and each voice can be tried for seven days. One detail matters if you work offline: while you are on trial voices, your data may be sent to Dreamtonics servers and offline mode is not available. Once a voice is bought and activated, Dreamtonics says all processing happens locally and your creative content is not uploaded.

Language support is six: English, Japanese, Korean, Mandarin, Cantonese and Spanish. Voices are native in five of those, and the cross-lingual synthesis feature lets any voice sing in any of the six. Inside the editor you get AI retakes of pitch, timbre and timing, vocal modes such as chest, belt and breathy, and a dedicated phoneme timing panel. It runs standalone on Windows and macOS or as a plugin in most DAWs.

Who it fits. Songwriters who want a demo vocal they can send to a singer, producers who dislike subscriptions, anyone who needs to work on a laptop with no connection. Dreamtonics says its official voices carry commercial use licences, and asks you to read each voice's licence for the details.

Who it does not fit. Anyone who needs French, Portuguese, Italian, German or Russian sung natively. And anyone expecting a wide stable of voices for $99: you get one, and each additional one is a separate purchase.

ACE Studio 2.0: the cloud workstation with more around the voice

ACE Studio describes its 2.0 release as an all-in-one AI music studio, and the product page backs that up. The Vocal Synth takes MIDI and lyrics: you draw or import the melody into a piano roll, click the text line under the notes and type the words. Pronunciation can be controlled down to the phoneme, and timing can be fine-tuned to the phrasing.

The language list is the longest of the three: English, Spanish, Italian, French, Portuguese, Mandarin, Japanese and Korean. The voice count is quoted as 140+ on some pages and 160+ on the pricing page, so take "well over a hundred" as the safe reading. Around the voice you get voice cloning slots (train a singing voice from your own recordings), AI instruments driven by MIDI, a stem splitter, a voice changer, and a feature called Inspire Me that the FAQ says can generate a song from a description or your own lyrics.

The trade-off sits in one FAQ line: Vocal Synth requires an internet connection for cloud-based vocal rendering. Generations usually take a few seconds, but no connection means no new vocal.

Pricing is plan-based, with rent-to-own, subscribe and buy options on the pricing page. At the time of writing the visible figures were for rent-to-own: the Artist plan at $16.58 a month billed yearly, and Artist Pro at $22 a month billed yearly, both described as "rent for 2 years, access forever". Artist includes 2,500 credits a month and one cloning slot for the vocal synth; Artist Pro doubles the credits and raises the cloning slots. The page also states a 7-day window to request a full refund. On commercial use, the FAQ says the pre-made voices and instruments are royalty-free for commercial projects, while a custom voice depends on the recordings you trained it on and your permissions to use them.

Who it fits. Producers who want many languages and a whole toolkit in one place, people who would rather pay monthly than a lump sum, anyone who wants to clone their own singing voice and then edit it note by note.

Who it does not fit. Anyone who works offline, and anyone who wants to own a licence outright with no plan running in the background.

VOCALOID: the character voices, and a new version one month away

VOCALOID is the oldest name here and the one most tied to specific characters. The VOCALOID home page features voicebanks such as HATSUNE MIKU V6 and lists older V3 and V4 banks alongside new ones. The current editor, VOCALOID6, is listed at $225 before tax on the purchase page. Its product page describes VOCALOID:AI, an AI engine for more expressive singing, more than 100 style presets, ARA2 support for DAW integration, and VOCALO CHANGER, which the page says imports vocal data to replicate your singing style as a VOCALOID voicebank. The free trial gives access to all of VOCALOID6's features.

Language support on the VOCALOID6 page reads Japanese, English and Chinese, with a single voicebank able to mix the three in one song. The VOCALOID shop also carries older voicebanks in other languages (the home page names a Spanish and a Korean one), but these are separate purchases, so check which editor version each one supports.

Here is the part that changes the buying decision. On 9 October 2026 Yamaha announced VOCALOID7, releasing on 11 November 2026. The announcement lists the editor at $129 before tax, bundled with a voicebank called "asa", an upgrade from VOCALOID3 to VOCALOID6 at $69 before tax (or $39 between 11 November and 20 December 2026), and confirms that all VOCALOID3 to VOCALOID6 voicebanks are compatible with the new editor. Buyers of VOCALOID6 products from 9 October onward get a $39 coupon toward the upgrade or a voicebank. What the announcement does not yet give is the feature list or the languages: Yamaha says those details will follow.

Who it fits. People who want a specific character voice and the culture around it, and anyone who already owns V3 to V6 voicebanks.

Who it does not fit right now. Someone starting from zero who is not in a hurry may want to watch the 11 November release first, since the new editor is listed cheaper than the old one. Run the VOCALOID6 trial in the meantime to learn whether the workflow suits you at all.

The gap none of the three closes: Russian and German

If your song is in Russian, none of these editors lists Russian on its own pages. German is missing from all three lists too. That matters, because a vocal synth pronounces through a language model of phonemes, and a language it does not know comes out as a foreign speaker guessing.

Some users try to spell Russian sounds with another language's phonemes. If you go that way, do it inside a trial before you pay, on a full verse rather than a single word, and expect an accent you will have to live with. For a Russian song where the voice matters more than note-level control, a song generator that takes lyrics directly is the more honest route; our guide to writing Russian lyrics for Suno covers what to watch there.

The part all three leave empty: melody and words

Each editor needs two things from you before it sings a note.

The melody you play on a MIDI keyboard, draw in the piano roll or import as a MIDI file. If you have the words and no tune, that is a separate problem, and the options are covered in can an AI melody generator give you a tune that fits your lyrics.

The words are where most people stall, and where a lyrics generator earns its place. Ours, at /generators, takes a short form brief (theme, emotion, structure, hook, point of view, language) and returns a title and a full lyric with section tags. It does not sing and never produces audio: the singing is what you bought the editor for.

Here is a lyric in the shape the generator returns, written for this page:

Title: Leave the Light On

[Verse]
The kettle clicks and no one comes to pour it
I count the cups we never put away
Your coat is on the hook, I keep ignoring it
The window holds the colour of the day

[Chorus]
Leave the light on
Leave the light on
I know the way back home
But leave the light on

Two things happen when this goes into a vocal synth.

First, the tags do not go into the piano roll. A note-based editor sings whatever sits under the notes, so [Verse] typed under a note would be sung as a word. Use the tags as a map instead: they tell you where each section of your melody starts, and how many lines it needs.

Second, every syllable needs a note. That is the actual fitting work, and it is where you will edit the lyric. Suppose the last verse phrase of your melody has eight notes. The line above has ten syllables:

melody phrase:  8 notes
original line:  The | win | dow | holds | the | col | our | of | the | day     -> 10 syllables
edited line:    The | win | dow | holds | the | last | of | day                 -> 8 syllables

"The window holds the last of day" keeps the image and lands one syllable per note. The chorus line "Leave the light on" is four syllables, so a four-note hook takes it as is. How a single syllable stretches across several notes, and how a two-syllable word is split, is handled differently in each editor, so check the manual of the one you picked rather than guessing.

Settings that help when the target is a vocal synth rather than a song generator: choose classic or loop structure so sections repeat cleanly over a repeating melody, a title-line hook so the chorus has one short line to land on, and set the vocal to female or male to match the voice you bought. Set the language field to the one your voice sings: of the generator's languages, en, es, ja, ko, fr, it and pt appear on at least one editor's list above; ru and de do not appear on any.

What it costs, plainly. One lyric on the SM1 model is 20 credits, which is $0.20. The SM2 and SM3 models cost 80 credits, $0.80, and run an extra editing pass: a detector flags weak lines and a repair step rewrites them. The monthly plan is $5 for 500 credits a month (25 SM1 lyrics), and plan credits do not roll over at renewal. The lifetime option is $19 once for 1,500 credits that never expire (75 SM1 lyrics); it is not unlimited. New accounts start at zero credits and there is no free trial of the generator. Prices are on /pricing. If you write your own words and enjoy it, you do not need us at all, and that is a perfectly good outcome.

Rights, briefly

Each vendor sets its own terms for its voices: Dreamtonics says its official voices include commercial use licences, ACE Studio says its pre-made voices are royalty-free for commercial projects and custom voices depend on your recordings. Read the licence of the specific voice or voicebank you buy before you release. On our side, generated lyrics may be used commercially, but they are non-exclusive and can overlap with what another user receives, and we clear nothing against anyone else's rights. This is not legal advice; our disclaimer states our position in full.

Pick in three questions

1. Does the song need Russian or German sung natively? None of the three lists either. Use a song generator for that one, or test phoneme spelling in a trial with low expectations.
2. Must it run offline, with no plan to keep paying? Synthesizer V Studio 2 Pro. One licence, one voice included, local rendering once the voice is bought.
3. Do you want eight languages, cloning and instruments in one place, and you are always online? ACE Studio 2.0. If instead you want a character voice or already own VOCALOID voicebanks, look at VOCALOID, and given the 11 November release, compare VOCALOID7's listed price before buying VOCALOID6.

Then do the smallest real test. Download the trial of the one you picked, draw eight bars of melody, generate one lyric with the structure set to loop and the hook to title-line, and type the chorus under your notes syllable by syllable. Twenty minutes with your own melody tells you more about the editor than any comparison, including this one.