Pick the narrowest tool that solves your actual problem
The most common mistake with v4.5 Plus is reaching for a transformation tool when a simpler operation would do. These three are powerful and they regenerate a lot of material - which means they can just as easily throw away the part you liked. Before you open any of them, work out what you're actually holding.
What you have decides the tool. Our own illustration.
| What you're holding | Reach for | Why |
|---|---|---|
| A finished instrumental, no vocal | Add Vocals | The arrangement already anchors the work - you only need the top line |
| A voice recording, no backing | Add Instrumental | The performance becomes the anchor and the band gets built around it |
| Several related Suno tracks | Inspire | The playlist supplies the shared traits, so you don't have to describe them |
| A song that's right but sounds wrong | Cover | Keeps the melody and identity - none of these three do that reliably |
| A song that's right but too short | Extend | Continues the recording you already have instead of making a new one |
| One section that isn't working | Song Editor | A local repair always beats a full regeneration |
| A voice you want across many songs | Personas | Add Vocals gives you a performance, not a reusable identity |
| Nothing but an idea | Custom mode | A strong style brief is the most direct route when there's no source audio |
Menu names and placement shift between releases - if a tool isn't where you expect, look under the track's own action menu rather than the Create screen.
Add Vocals
Takes an instrumental and sings over it. You supply lyrics and a style description; Suno works out the melody, the phrasing and where the lines sit. It also does vocal replacement - feed it a full song and you can swap the voice rather than add one.
Instrumental in, finished song out
Best when the arrangement, groove and harmony are already what you want and the only missing element is the voice. Worst when you're secretly hoping it'll fix the instrumental too - it won't.
- A clean instrumental - uploaded, generated in Suno, or pulled from a stem export
- Lyrics that are close to final, with readable section labels
- A style description of the finished result, not just the vocal
- Check the arrangement actually leaves space for a vocal - a dense mix has nowhere to put one
- Write lyrics that match the section lengths and the energy of each part
- Open Add Vocals from the track and describe the whole vision in the style field
- Test how closely it's following the source before rewriting anything
- Judge phrasing and hook clarity first, then whether the arrangement survived
- The vocal doesn't sit naturally. Usually lyric density against instrumental spacing - cut words, not instruments
- It drifts off the source. Adjust the source-influence control before touching the prompt
- The lyrics won't fit. Simplify the style brief first; changing the source material is the last resort
Add Vocals is not a voice identity. It gives you a performance on one track. If you want the same singer across a body of work, that's Personas - and using Add Vocals repeatedly and hoping for consistency is the single most common way people waste credits on this feature.
Add Instrumental
The mirror image. Give it a voice - an a cappella, a vocal sketch, a hummed melody, an isolated stem - and it builds an arrangement around it. It will also swap an existing backing track for a different one, which is how people move a song between genres without regenerating the vocal.
Voice in, arranged track out
Best when you have a performance worth keeping and no way to produce around it. Suno's framing is that it lowers the barrier for people without recording equipment - though you still need a usable recording to start from.
- The cleanest vocal you have - room noise and bleed get baked into the result
- One clear genre lane, not three
- Tempo language if you know it, even loosely - "unhurried", "driving"
- A short instrument hierarchy: what leads, what supports
- Work out the tempo feel, the emotional character and the likely song form before you type anything
- Pick one primary genre and name the instruments in priority order
- Generate, then listen specifically for whether the arrangement fights the vocal rhythm
- Reject anything that does - that's a structural mismatch, not something a reroll fixes
- Decide whether the failure is local (one section) or global (wrong genre lane)
- The band crowds the vocal. Ask for fewer featured instruments and a supportive arrangement
- The arrangement wanders. Tighten the genre and be more specific about tempo
- It sounds glued together. Almost always the source - a noisy or clipped vocal never recovers
There's a large overlap here with the broader audio inputs feature, and the distinction is worth knowing: audio inputs use your recording as a starting point the model can depart from, while Add Instrumental treats the vocal as fixed and builds to fit it. If you want your melody preserved exactly, this is the one.
Inspire
The odd one out. Instead of transforming audio, Inspire reads a playlist of your own Suno tracks, works out what they have in common, and writes something new that belongs with them. Think of it as describing a style by example rather than in words.
Playlist in, a track that fits out
Best for continuing a lane you've already established - a series, a project, a set of tracks for one video channel. Useless if the playlist is just "songs I liked".
- A tightly focused playlist of your own Suno songs - not your whole library
- Tracks that genuinely share an identity, not just a mood
- New lyrics, and a clear idea of what this song is for
- Build the playlist deliberately - include only songs that belong together creatively
- Remove outliers ruthlessly; one wrong track dilutes the whole reference
- Give it new lyrics and state the specific objective for this song
- Judge the result on shared identity, not on whether it resembles any one track
- Note which traits carried through - that tells you what the playlist is actually communicating
- Generic results. The playlist is too broad. Cut it down until every track earns its place
- Too much like the references. Say explicitly what the new song must contribute that they don't
- Nothing carries through. The tracks share less than you thought - listen back and check
Inspire and Personas answer neighbouring questions. A Persona locks a voice and style so successive songs sound like one artist. Inspire is looser - it carries mood, tempo and instrumentation without pinning the vocal. If consistency matters, use a Persona; if you want family resemblance with room to move, use Inspire.
Source material is the whole game
One rule governs all three tools, and it's unglamorous: they cannot rescue a bad source. Add Instrumental on a vocal recorded in a kitchen produces a produced track with a kitchen in it. Add Vocals over a mix with no headroom produces a vocal fighting for space. The preparation matters more than the prompt.
If you're uploading rather than using something you generated in Suno, the limits are modest:
Upload capability expanded during 2025 and these figures come from Suno's help centre rather than the app - treat the uploader itself as authoritative if it offers you more.
What a good source looks like
- Dry. No reverb baked in, no room tone. The model will add ambience; it can't remove it.
- Level and uncompressed. Clipping and heavy limiting both confuse the analysis.
- One thing at a time. An isolated vocal, not a rough mix with a vocal in it.
- Musically clear. A steady tempo and an obvious key give the model something to build on.
What will waste your credits
- Phone recordings with background noise. Every artefact survives into the finished track.
- Material you don't own. You have to confirm you do, and the August 2026 guidelines are explicit about voices and likenesses.
- A dense instrumental with no vocal space. Add Vocals has nowhere to put the line.
- A wandering tempo. Add Instrumental needs something to lock to.
Each tool locks one element and regenerates the rest. Knowing which is which is most of the skill.
Suno v4.5 Plus FAQ
What is v4.5 Plus?
An incremental release on 18 July 2025 that kept the v4.5 model and added three production tools - Add Vocals, Add Instrumental and Inspire - for Pro and Premier subscribers. The model number barely moved; the toolkit did.
Can Add Vocals replace the singer on an existing song?
Yes - that's vocal replacement, and it's one of the two things the tool does. Feed it a full song rather than an instrumental and you can swap the voice. The usual caveat applies about material you don't own.
Can I use my own singing with Add Instrumental?
That's exactly what it's for. An a cappella, a vocal sketch, a hummed melody or an isolated stem all work - the cleaner the recording, the better the result, because nothing in the process removes room noise.
How long can my upload be?
Suno's help centre gives 6 to 60 seconds on the free plan and up to 120 seconds for Pro and Premier. Upload capability has expanded over time, so trust the uploader in the app if it offers more.
Do these work on the free plan?
They shipped to Pro and Premier. Free accounts get a much shorter upload limit and don't have the production tooling, so this is a paid-tier feature set in practice.
Add Vocals or Personas?
Different jobs. Add Vocals puts a performance on one track. Personas save a voice and style so a whole catalogue holds together. Running Add Vocals repeatedly and hoping the singer stays the same is the classic mistake.
Inspire or a Persona?
A Persona locks the voice and style. Inspire is looser - it carries mood, tempo and instrumentation from a playlist without pinning the vocal. Use a Persona for consistency, Inspire for family resemblance.
Why is my Inspire result so generic?
Almost always the playlist. If the tracks only share a vague mood, the only thing they have in common is "pleasant", and that's what you'll get. Cut it down until every song in it earns its place.
The new vocal doesn't sit properly. What now?
Look at lyric density against the instrumental before you touch anything else. Too many syllables for the space is the most common cause, and the fix is cutting words rather than thinning the arrangement.
Are these still available in 2026?
These are features rather than a model, so they're not covered by model retirement - but placement changes between releases and Studio has absorbed some of this territory. Check the track's own action menu if a tool isn't where you expect it.
Sources
We're an independent guide with no affiliation to Suno, Inc. Menu placement and tool naming change between releases - where our steps and the app disagree, the app is right. Upload limits are quoted from Suno's help centre and have expanded over time. Only upload material you have the right to use; Suno requires you to confirm this, and the August 2026 community guidelines are explicit about voices and likenesses. The diagrams are our own illustrations. Last reviewed August 2026.
Prepare the source, then pick the tool
A clean, dry, isolated recording and the narrowest tool for the job will beat any amount of prompt rewriting. These features reward preparation more than anything else in Suno.