Suno AI Voice Persona: How to Clone Your Voice With Suno's New Native Workflow
Suno AI Voice Persona: How to Clone Your Voice With Suno's New Native Workflow
Who is @creativelyMakeAIMusic?
CreativelyMakeAIMusic is an educational YouTube channel that teaches musicians and songwriters how to create original music using AI tools, songwriting techniques, and creative workflows for modern music production, including platforms such as Suno. The channel provides tutorials on AI music production, vocal control, songwriting techniques, and creative workflows using platforms such as Suno. Its content helps adult musicians and creators learn how to integrate AI into modern music creation and production.
What if you could sing a quick line into Suno, then hear it come back as a polished version of your own voice, ready to use in a full song? That's the promise behind a Suno AI voice persona, and the newest workflow makes it feel less like a workaround and more like something Suno almost meant you to do.
This post walks through a real test of Suno's native process for turning a short vocal recording into a reusable persona. You'll see the exact flow, from recording a sample, to choosing the best clip, to dialing in settings like audio influence, and finally using that persona inside a new creation. Along the way, you'll also get an honest sense of what worked, what didn't, and how close the results can get right now.
If you want to follow along in your own account, use the Suno sign-up page shared by Creatively Make AI Music.
What a Suno voice persona is (and why people want it)
A Suno voice persona is a saved voice identity you can reuse in future generations. Instead of prompting for a generic vocalist every time, you feed Suno a short recording, then keep that "version" of the voice available for new songs.
That idea matters because vocals are personal. Even a simple melody can feel like a diary entry once a familiar voice sings it. For creators who write their own lyrics, a voice persona can also help keep songs consistent across a project. The same "singer" shows up again and again, which makes a set of tracks feel connected.
Just as important, this test focuses on doing it inside Suno, without the extra steps some older workflows needed. The video description calls out tools people often used for earlier approaches, like Controlla Voice, LALAL, or Kits. In this newer flow, the goal is to record directly in Suno, generate results, then turn the best output into a persona without jumping through hoops.
The big shift is simple: record in Suno, create a sample, then save it as a persona from the same workspace.
That doesn't mean everything comes out perfect on the first try. As you'll see, one attempt missed the mark, while the next one landed closer to an actual "me singing" result.
Record your singing inside Suno and save it to your library
The process starts with a short vocal recording. In Suno, the first step is to go to the audio area and use the record option. The test used a short sample of singing, nothing long or complex, because the point was to see how well Suno could pick up the vocal character.
After recording, you choose the portion you want Suno to use. Think of this like selecting the best few seconds of a take. You're not trying to impress anyone with range or power here. You're trying to give the system a clean slice of your voice that it can actually read.
Once that segment is selected, the audio gets saved to the library. After processing, it appears in the workspace. At that point, you can drag the audio into the create screen and drop it in.
This part feels like setting up a paint palette before you start. If the sample is messy, the painting will be messy too. If the sample is clear, the results have a better chance.
A detail that matters here is that Suno can pick up lyrics from the recording. In this test, the line being sung was essentially, "I am going to test my voice in this software." Suno recognized that lyric content and used it as part of the creation step. That lyric recognition can be helpful, but it also shows why a clean recording matters. If your words blur together, the model has to guess.
Choose the right sample section and dial in the settings
Once the audio is in the create screen, Suno lets you choose "sample" and then pick the section you want to use as the basis. This is where you lock in the snippet that will guide the voice.
Next comes the text prompt side. In the test, a description was typed in, and some styles were excluded. Then the key slider got pushed hard: audio influence was set to around 90 percent.
That number wasn't random. The test notes that pushing it too far can lead to "red weirdness," where the output starts to get strange. So 90 percent sat close to the edge, strong enough to keep the voice character, but not so extreme that it breaks.
Meanwhile, style influence and weirdness were turned way down. In other words, the goal was not to create something wild. The goal was to hear the voice, as clearly as possible, in a Suno-generated result.
Here's a quick snapshot of the settings used in this test, based on what was stated:
Setting What was done Why it matters
Sample selection Picked a specific vocal section Gives Suno the cleanest voice reference
Audio influence Set around 90 percent Keeps the voice identity strong
Style influence Turned way down Reduces stylistic drift away from the real voice
Weirdness Turned way down Helps avoid unpredictable artifacts
Style exclusions Excluded a couple of styles Narrows the target so the model behaves
The takeaway is simple: the more you want "you," the more audio influence matters. At the same time, the more you push the system, the more it can glitch. This test aimed right at that tension.
Compare the first results, then pick the best take
After hitting create, Suno generated versions of the voice sample. The first one did not work well. The second one landed better and produced a result that felt like it actually "made me sing."
The output included the same basic lyric line, with music behind it. The test also created two more versions to see if results improved, but the earlier stronger take became the one chosen to turn into a persona.
This is an important mindset shift if you're used to thinking in single outputs. Treat it more like recording takes. One take can fall flat. Another can surprise you. Suno's generation step becomes a kind of audition process where you listen for the one that carries your tone in a believable way.
A short line is enough to test this. You don't need a full chorus to answer the main question: does it sound like you?
In the sample shown, Suno repeated the lyric in a musical phrase, essentially singing, "I am going to test my voice in this software." Even with that simple text, you can hear the difference between a result that misses and one that clicks.
If you want another angle on using your voice creatively in Suno, the channel also shared a related project focused on using your voice as a prompt. Here's the "make songs using your voice as a prompt for Suno" video.
Save the sample as a Suno AI voice persona (the new workflow change)
The biggest workflow change described in the test is what happens next. Instead of exporting stems or doing extra prep, you can now make a persona directly from the result.
The flow goes like this: click on the three dots for the created item, go to create, then choose the option to make a persona. That's the moment where the experiment stops being a one-off and becomes reusable.
Once the persona exists, you can head back to the create screen, put the persona in, and build a song around it. In this test, the lyrics were already there, then a genre got added to the style box. Audio influence was turned up again to keep the persona strong in the final output.
This "persona to song" step is where the workflow starts to feel practical. You are no longer stuck with a single clip. You can take that voice identity and try it in different styles, different tempos, or different lyric structures.
It also lowers the barrier for experimentation. When creating a persona takes fewer steps, you're more likely to test variations, keep what works, and scrap what doesn't. That matters because voice is finicky. A tiny change in the sample can change the feel of everything that follows.
What the results sounded like, and what that means for creators
The conclusion of the test was straightforward: Suno turned the creator's voice into a clone voice and, in their words, it "turned me into a better singer." That's a big claim, but it matches what many people hope for when they try voice-based AI tools. They want to keep the identity while smoothing out the rough edges.
At the same time, the test didn't pretend the output was flawless. One of the early generations didn't work well. That detail matters because it sets expectations. You may need multiple attempts, even with the same input, to get a result you'd actually keep.
The most useful part of this test is the way it isolates variables:
A short, direct sung phrase, recorded inside Suno
A selected sample section, not the whole recording
Audio influence pushed high (around 90 percent) without crossing into the "red weirdness" zone
Style influence and weirdness turned way down to avoid drifting away from the voice
That combination aims at one thing: voice similarity over everything else.
If the goal is "use my voice," then the settings should protect the voice first, and decoration should come second.
For songwriters, this opens a door. You can sketch lyrics, hum a melody, record a small snippet, then see if Suno can carry your vocal identity into a full arrangement. Even when the result is not a perfect match, it can still be a powerful writing mirror, reflecting your idea back with new energy.
Conclusion: A simpler path to a Suno AI voice persona
This native workflow makes a Suno AI voice persona feel much more reachable: record a short vocal, choose a clean section, generate a few options with strong audio influence, then save the best one as a persona and use it in a song. The test also shows the honest truth, not every generation hits, so it pays to try a few.
If you try the same process, compare your first and second results closely, then build from the best take. After that, the real fun starts when you drop the persona into new genres and see how your voice holds up. Thanks for reading, and if you run your own test, share what surprised you most.
