ACE Studio Review (October 2026): AI Vocal Synthesis for Producers Without a Singer

ACE Studio turns a melody and a line of lyrics into a sung vocal. You draw or play the notes, type the words, pick a singer from a library of more than 160 voices, and the software renders a performance you can edit note by note and drop into your song. Version 2.1.8, released on September 17, 2026, is current as we write in October 2026.

This review covers what the program does, what it costs, what you may release, and how it compares with Synthesizer V and VOCALOID. Every feature, price and licensing term below comes from ACE Studio's own documentation, pricing page and Terms of Use, read between October 11 and 16, 2026, and each fact is dated. To show the editing in musical terms, we take one public-domain hymn, "Amazing Grace," through the whole process: syllables, breaths, dynamics, vibrato and a second harmony part.

ACE Studio 2 in six facts, as listed on its own pages in October 2026. "From $398" is the 20 percent holiday price shown on October 16 (list $497.50); prices change often, so check the pricing page before you buy.

What ACE Studio 2 is in October 2026

ACE Studio is made by Timedomain, which announced version 2.0 on December 8, 2025 as an "all-in-one AI music studio." The launch added a new vocal model (Verse 2), more than 140 voices, four new singing languages for a total of eight, AI string, brass, saxophone and duduk instruments, a redesigned arrangement view called the Canvas, and Turbo Mode for faster rendering on your own graphics card.

Updates have come every few weeks since then. The public changelog lists seventeen releases between 2.0.0 and 2.1.8. The ones that matter most to a singer-less producer: third-party VST and AU effects inside the app (2.1.2, June 2026), four new fine-tuning parameters for V2 voices (2.1.6, August), editable phoneme boundaries and a simpler Dynamic control (2.1.7, September), and third-party VSTi instruments plus command-line and MCP control for AI agents (2.1.8). The voice library has grown to "160+" on the pricing page.

Ten months of updates, from ACE Studio's public changelog, checked on October 16, 2026. Version 2.1.8 of September 17 was still the newest.

The program runs on Windows 10 or 11 (version 22H2 or later) and macOS 13 Ventura or later, and the installation page is blunt about one thing: "ACE Studio requires an Internet connection to run." That single fact separates it from both competitors further down.

How a vocal gets made

The workflow has six stages. You enter notes: draw them in the piano roll, play them in from a MIDI keyboard, or import a MIDI or MusicXML file. A vocal clip is monophonic, so one clip holds one melody line. You type lyrics, one syllable per note. ACE converts the words into phonemes, the individual speech sounds, according to each note's language setting, and you can correct any word it gets wrong. Then you choose a voice and the app renders.

The six stages of an ACE Studio vocal, drawn from its documentation. You will loop back to stages 1 to 3 after every listen.

Rendering happens in the cloud by default, in pieces of about a phrase, and only the part you changed is rendered again, so small edits come back quickly. Turbo Mode moves the heavy steps onto your own graphics card if it is supported: on Windows an NVIDIA RTX 2060 or newer, an AMD RX 6600 XT or RX 7600 and up, or Intel Arc; on a Mac an M1 Max, or any M2 or newer. If your machine fails the device test, ACE falls back to the cloud.

When the vocal is done you export audio, or you keep working inside your DAW (digital audio workstation) with the ACE Bridge 2 plugin, a VST3, AU and AAX instrument that sends your DAW's MIDI to a linked ACE track and streams the rendered voice back. Tempo and time-signature sync uses a separate ACE Bridge ARA plugin.

The gear a vocal synth stands in for: a condenser microphone and a pop filter. Photo: Dejan Krsmanovic, CC BY 2.0, via Wikimedia Commons (cropped).

A test piece: "Amazing Grace" from a real hymnal

Any melody shows the controls, but a melody with known words and a printed score shows whether the result is musical. We used "Amazing Grace" (words by John Newton, 1779; tune NEW BRITAIN, 1829, which this hymnal labels "McIntosh"), copied note by note from Hymns of the Faith, edited by E. O. Excell and A. C. Smither (St. Louis: Christian Board of Publication, c. 1918), No. 87, scanned at the Internet Archive. It is in G major and 3/4 time and starts with a pickup, a single beat before the first bar line; we number that pickup bar 0. Everything here is public domain, so you can type it in and publish the result.

The melody and first stanza as printed in Hymns of the Faith (c. 1918), No. 87: a one-beat pickup and fourteen bars, with slurs only on "once" and "now," as on the page. The examples below all come from this page.

Put the rhythm on the grid first

A vocal synth sings exactly what is on the grid, so count the rhythm before you draw it. The pickup "A-" starts on beat 3. In bar 8, "once" starts on a dotted quarter note, a beat and a half, and its second note falls halfway through beat 2. ACE's grid can be set from 1 to 16 cells per beat, and the hymn needs only two: eighth notes. If the counting is new to you, Pickup Notes explains the upbeat start, and The Dotted Quarter Note drills that long-short rhythm.

Counts under the notes: the pickup starts on beat 3, and the dotted "once" in bar 8 pushes the next note to the "and" of beat 2.

Syllabic and melismatic setting

Most of the hymn is syllabic: one note per syllable, as in "grace! how sweet the sound." Seven places are melismatic, one syllable stretched over two notes, such as "once," "was," "now" and "am" in bars 8 to 10. Modern editions mark a melisma with a slur and an extender line after the syllable; this hymnal often just beams the two eighth notes together.

ACE Studio handles this with one rule from its lyrics documentation: every note takes one syllable, and a note with a dash "-" instead of a syllable continues the previous one. The Note Brush and Note Slicer tools make the same kind of continuation note when you draw inside a note or split it. If you type "once was" across four notes, the voice sings four syllables and the line falls apart; type "once - was -" and it sings two.

Top lyric line: the words as sung. Bottom line: what you type into ACE Studio, one entry per note, with a dash wherever a note continues the syllable before it. The slurs on "was" and "am" and the extender lines are editorial; the hymnal slurs only "once" and "now."

The Ties vs. Slurs sheet is a good check here: a slur over two different pitches on one syllable is a melisma, while a tie over the same pitch is just one longer note, which you should draw as one long block.

Breaths

A real singer breathes between phrases, and a synth that never breathes sounds like a machine. In the V2 voices the first tab under the editor is Breath & Phoneme. Its "Add all" button places breaths for the whole clip, mostly at phrase starts; you can also click before a phrase to add one and drag its left edge to make it longer. The documentation's own tip is to put a breath before long lines and high notes.

Breath marks (editorial; the print has none) after "sound," and "me!": shorten each long note slightly and breathe before the next pickup.

Where to breathe is a phrasing decision, and the hymn makes it easy: each line of the poem is four bars long and ends on a long note. The Phrasing sheet uses exactly this mark, the comma, and the idea of rising through a phrase and tapering at its end. Phrase and Period shows why four-bar phrases pair up the way they do.

The same tab splits every note into its phonemes. Since 2.1.7 you can drag a boundary to lengthen a consonant or a vowel, and an adjusted phoneme turns blue and stays locked.

The three tabs under a V2 vocal clip, drawn as a schematic, not a screenshot. Tab and parameter names follow ACE Studio's Vocal Parameters page (October 2026).

Dynamics

The Dynamic tab holds one envelope that runs from soft to powerful. Left undrawn, it stays on Auto and the AI decides. Draw it with the brush, or use the Line tool for an even crescendo or diminuendo. Version 2.1.7 replaced the older set of Power, Soft, Breathy and Chest controls with this single curve.

One dynamic plan for the second half of the verse (editorial): mf at the top of the tune on "once," fading to mp by "found," then p, a small swell toward "now," and pp on "see."

Write your dynamic plan on the score first, then draw the envelope to match it. The top of the tune, D5 in bars 7 and 8, is the natural loudest point; the last line can start softer and fade out. The Dynamic Ladder sheet lays out the six steps from pp to ff that we used. The Advanced tab adds four finer curves, Air, Falsetto, Tension and Energy, each with values from 0 to 1 (the 2.1.6 changelog calls the first one Breath). The docs advise small changes, and that is good advice: a big Tension value on every note is the vocal equivalent of shouting.

Pitch and vibrato

ACE draws two pitch curves. The AI pitch, shown dark, is what the voice does by itself: scoops into notes, small drifts, and vibrato on many long notes. Your user pitch, shown white, overrides it wherever you draw. Six tools edit it: Pitch Brush, Pitch Anchor, Fixed Brush, Pitch Eraser, Vibrato and Modulation. The Vibrato tool adds a vibrato with handles for amplitude (how wide), frequency (how fast), phase, and attack and release (how it fades in and out). Modulation flattens or exaggerates the existing movement.

One sung half note on "grace!" as a pitch curve: a scoop, a straight start, then vibrato. A drawing of the idea, not a screenshot.

Placement matters more than depth. Singers in this style tend to keep short notes straight and let vibrato bloom on held notes at the ends of lines. The documentation warns that adding your vibrato on top of the AI's makes the pitch chaotic, and suggests drawing a flat user pitch first, then adding vibrato. For a choir sound, its choir guide suggests flattening vibrato with the Modulation tool.

Vibrato placement (editorial): vibrato on the line-ending "found," and "see." and nearly straight tone on "blind," and "now," so the ending sounds like an arrival.

A second part for harmony

Because a vocal clip is monophonic, each harmony part needs its own track. The hymnal prints an alto part under the melody, and it makes an ideal first harmony: it sings the same words in the same rhythm, with one note per syllable, and in bars 0 to 7 it uses only three pitches, B3, C4 and D4.

Melody and alto in bars 0 to 7, as printed in Hymns of the Faith (c. 1918). Give the alto its own track and its own voice; it sings every syllable on one note.

Two related tools are easy to confuse. "Create vocal doubles" makes copies of a part with small random changes in timing and consonant length; the docs suggest a similarity of 85 to 90 percent and explain why a plain duplicate fails: identical audio causes phase problems. Unison Mode (called Choir Mode up to version 2.1.1) puts several voices on one track singing the same line, with Offset, Spread and Gain settings. Neither one writes a harmony for you. If you want to write your own, the Four-Part Harmony sheet shows how soprano and alto share a staff and stay close together.

A harmony stack for bars 0 to 4, drawn from the printed notes: lead, alto and two doubles, each on its own track.

Put together, the steps above make a short routine you can repeat with any tune: rhythm and lyrics first, then breaths, then dynamics, and pitch work last.

A first-hour plan with any vocal synth, in the order used above. Vibrato goes last because pitch edits depend on everything before them.

Voices, languages and cloning

The library holds "160+" royalty-free pre-made voices and "40+" AI instruments, according to the pricing page. Current voices use the Verse 2 model family, Verse 2.7 in October 2026, which sings in English, Spanish, Italian, French, Portuguese, Korean, Chinese and Japanese. Every note has its own language setting, so one line can switch languages, but the lyrics are not translated for you, and each voice still sounds best in its native language. ACE also imports MusicXML, and its choir-rehearsal guide even suggests voices by name for each choir part.

Voice cloning trains a model on your own singing. The guidance is 30 to 100 minutes of clean, dry recordings, with little gain past 120 minutes: no reverb, no background noise, no harmonies or doubled takes, and no phone, laptop or headset microphones. The cloned voice is private to your account. Cloning someone else's voice is where the rules tighten; see the licensing section below.

A real vocal session. Training a cloned voice asks for the same thing a good session gives you: clean, dry takes. Photo: Laurahuet, CC BY-SA 3.0, via Wikimedia Commons (cropped).

Plans and prices

ACE Studio 2 has two tiers, Artist and Artist Pro, and two ways to pay. The pricing page lists Artist at $497.50 to buy outright or $20.73 a month billed yearly as "rent-to-own," and Artist Pro at $660 or $27.50 a month. On October 16, 2026 it also showed a 20 percent "holiday discount" on all of these: Artist at $398 or $16.58 a month, and Artist Pro at $528 or $22 a month. Rent-to-own turns into a lifetime license after two years, and two years of payments come to about the purchase price, at list or discounted prices.

Artist vs. Artist Pro as listed on October 16, 2026: list prices in US dollars before tax, with the 20 percent holiday discount then on offer. The main differences are cloning slots and monthly credits.

Both tiers include unlimited use of the pre-made voices and instruments. The differences are cloning slots, 1 vocal and 1 voice-changer slot on Artist against 5 and 10 on Artist Pro, and monthly credits, 2,500 against 5,000. Credits pay for the extras: the generative music and SFX kits, Music Video, Video Composer and the Stem Splitter. If you stop paying before you own a perpetual license, cloned voices on subscription slots are locked until you resume, and credits stop refreshing after the current period. A 7-day money-back guarantee is offered by email. Full-time students and teachers can apply for a 30 percent discount code with proof of status or an .edu address.

What you may release

The short answer is that you can release songs made with the pre-made voices. ACE Studio's Terms of Use (effective August 11, 2026) say you keep the rights to the audio you create, that ACE claims no copyright in it, and that you may exploit it commercially. Those rights survive cancellation for anything made while your membership was active. ACE also commits not to train AI models on your uploads or output.

Licensing in one picture, summarized from ACE Studio's Terms of Use, License & Copyright page and custom-voice guidelines (October 2026). Not legal advice.

There are three catches. Not every pre-made voice is license-free: the Terms say "the majority are," and a few voices carry a "third-party license required" tag, so check the voice profile, where you can also generate a PDF license certificate. A cloned voice works only inside ACE Studio, so it needs a membership. And cloning another person needs their consent. ACE's custom-voice guidelines say it may ask for written permission and a recording of verbal consent, and may end the service if you cannot provide them. Simulating politicians or government officials is banned even with their consent.

ACE Studio vs. Synthesizer V and VOCALOID

The two older names in vocal synthesis are both one-time purchases. On October 16, 2026, Dreamtonics' store listed Synthesizer V Studio 2 Pro at $99 with one voice of your choice, and extra voices at $79 each. It renders on your own computer once a voice is activated, runs as a standalone app and as a VST3, AU, AAX and ARA plugin on Windows and macOS, and sings six languages. Updates are free within version 2; the latest update announced on Dreamtonics' blog is 2.2.1 (March 2026), and a 2.3.0 beta, posted on its forum on October 8, 2026, is testing more languages, French among them. Dreamtonics says its official voices include commercial-use licenses, subject to each voice's terms.

Yamaha's VOCALOID6 Editor is $225 with 25 voicebanks included. On October 9, 2026 Yamaha announced VOCALOID7 for November 11, 2026: the editor at $129, an upgrade at $69 ($39 until December 20), and compatibility with all VOCALOID3 to 6 voicebanks. Yamaha's FAQ says songs made with its own VOCALOID6 voicebanks can be sold without a license agreement "in principle," with each voicebank's license to check.

The three vocal synths side by side, from each maker's pages (October 11 to 16, 2026). VOCALOID7 was announced but not yet released at the time of writing.

The practical difference is not sound quality, which you should judge with your own ears in the free trials and the money-back period. It is the business model. ACE gives you a large library and cloud tools for a bigger payment or a two-year rental, and it needs the internet. Synthesizer V costs least to start and works offline, but each extra voice is another purchase. VOCALOID sits between them, and anyone planning a purchase now should look at the VOCALOID7 upgrade terms first.

Pros and cons

The trade-offs in eight lines. "Costs more up front" compares the one-time prices of the three products in the table above.

ACE Studio is a strong fit if you write songs and need a believable demo singer, a harmony stack, or a sketch in a language you do not sing in, and you are happy to pay for a large library and work online. It is a weaker fit if you need to work offline, want the lowest entry price, or only ever need one voice.

What a vocal synth will not do for you

Every musical decision in this review came from reading the score: where the pickup falls, which syllables stretch, where to breathe, where the line peaks, which notes get vibrato. ACE Studio will sing whatever you enter, and the AI fills in a plausible performance, but it cannot tell you that "once" starts on a dotted quarter or that the alto should stay below the melody.

Choir rehearsal in Brecon Cathedral. Breathing, phrasing and blending are what a synth imitates, and what live singers practice every week. Photo: Andy Dingley, CC BY-SA 3.0, via Wikimedia Commons (cropped).

If you want to read melodies, phrasing and harmony more confidently, a Tissimo voice teacher can work through a song like this with you in a live online lesson, and check the breaths and dynamics you plan before you draw them.

Free worksheets for this lesson

• Pickup Notes — how a tune starts before the first bar line, with "Amazing Grace" among its examples.
• The Dotted Quarter Note — the long-short rhythm of "once" in bar 8.
• Ties vs. Slurs — tell a held note from a slurred melisma.
• Phrasing — phrase marks, the comma breath mark and shaping a line.
• Phrase and Period — four-bar phrases, questions and answers, cadences.
• Dynamic Ladder — the six dynamic levels from pp to ff.
• Four-Part Harmony — soprano, alto, tenor and bass, and how the parts move.
• Reading a Lead Sheet — "Amazing Grace" in C with chord symbols, a quick way to build a backing track.

Sources

• Pricing — ACE Studio
• Terms of Use (effective August 11, 2026) — ACE Studio
• ACE Studio 2.0 changelog — ACE Studio Docs
• Vocal Parameters — ACE Studio Docs
• Editing Pitch — ACE Studio Docs
• Editing Lyrics — ACE Studio Docs
• License & Copyright — ACE Studio Docs
• Usage Guidelines for ACE Custom Voice Model — ACE Studio Support
• Synthesizer V Studio 2 Pro — Dreamtonics
• VOCALOID7 Coming November 11, 2026 — Yamaha VOCALOID
• Selling songs made with VOCALOID6 voicebanks (FAQ) — Yamaha VOCALOID
• Installation — ACE Studio Docs
• Local Rendering — ACE Studio Docs
• Cloning a Voice for Vocal Synth — ACE Studio Docs
• ACE Bridge 2 — ACE Studio Docs
• ACE Studio 2.0 Released (December 8, 2025) — ACE Studio Blog
• Synthesizer V Studio 2 Pro bundle (prices captured October 16, 2026) — Dreamtonics Store
• Synthesizer V Studio 2 Pro 2.2.1 — Dreamtonics
• Synthesizer V Studio 2 Pro 2.3.0 Beta 1 (October 8, 2026) — Dreamtonics Users Forum
• Purchase VOCALOID6 Editor — Yamaha VOCALOID
• Hymns of the Faith (c. 1918), ed. Excell and Smither — Internet Archive