Super Mode Now Automatic + Smooth Pass Retired: What Changed

Quick update on this post. Super Mode is now applied automatically on every song in the ChangeLyric dashboard. There is no checkbox to toggle anymore. Every job gets the deeper analysis by default.

We also retired Smooth Pass. Quality was not where we wanted it, and we would rather kill a feature than ship something that disappoints you. The rest of this post explains what Super Mode does and what to use now in place of Smooth Pass.

What Super Mode Does

When you submit a song, the AI now spends several minutes analyzing your uploaded track before generating anything. It studies the vocal characteristics, how the singer transitions between phrases, and how the voice sits in the mix.

Then when it generates your new lyrics, it has that deeper context to work with. The result is vocals that match the original more closely: better tone matching, smoother transitions between changed and unchanged sections, and output that preserves the feel of the original recording.

The Time Tradeoff

The first time we run on a new song, the analysis adds roughly 5-10 minutes of processing time. That is the AI doing its deep work upfront.

But that analysis is cached. Every subsequent edit on the same song, whether different lyrics, different sections, or new versions, runs at normal speed. The AI already knows the song, so the real cost is a one-time wait per song.

If you are doing multiple iterations on a song (which you should be), the upfront analysis pays for itself after the first edit. It is included with your plan at no extra cost.

How To Use It

Just upload your song in the dashboard and submit. Super Mode is on by default for every job, so there is nothing to toggle. Your workflow is the same as before.

Smooth Pass Has Been Retired

We previously shipped a Beta refinement tool called Smooth Pass that let you build a scratch track in your DAW and return a polished isolated vocal. It is no longer available. Output quality was not consistent enough to keep it in the product.

Rather than keep a feature that disappointed people, we pulled it. Below is what to use instead, plus what we are building next.

Use The Voice Changer To Refine Output

If your Super Mode result is close but you want to push tone matching further, run the vocal through the voice changer. You get two paths there.

Option 1: Pick a curated model. We have over 200 curated voice models covering a wide range of genres and vocal styles. Find one that matches the original artist closely and run your output through it for a tighter tone match.

Option 2: Train your own model. If you need the best possible match for a specific song, train a custom model on the original vocal. A model trained on the exact track you are editing will outperform any generic model for that song.

See our guide on the best way to change a vocalist with AI for the full workflow.

Vocal Upscaling Coming To V3

We are working on integrating vocal upscaling, powered by the same voice changer pipeline, directly into the V3 tool. The goal is one-click refinement on every job, no separate step required. Expect this to land soon.

A Note On Using Suno For Smoothing

If you have used Suno before, their refinement step is genuinely good for smoothing rough vocal tracks. It is a legitimate option in your toolkit, not a competitor we are pretending does not exist.

One catch: Suno's content moderation is strict. If you upload a track that still has the original instrumental on it, their system can misdetect it as copyrighted material and block the job. The workaround is to remove the instrumental from your input audio before sending it to Suno, so they only see the vocal you are trying to refine.

You can pull a clean vocal stem out of any mix using the stem separation step that already runs as part of your ChangeLyric job, then take that isolated vocal into Suno for smoothing.