작곡가가 Suno 음질 3배 올리는 비밀 공식 (학습 음원 무료 배포) — Transcript
Full transcript
- 0:11Did you know you can create your own
- 0:13custom model in Suno? With just this
- 0:15one thing, you can completely get rid
- 0:16of that characteristic AI tinny noise.
- 0:18The details in the muffled sound come
- 0:20alive, and the sound that was trapped
- 0:21in the center spreads out wide. "Ah,
- 0:23the melody and lyrics are great, but
- 0:24something feels like it's missing."
- 0:26That's highly likely to be a sound
- 0:27issue. And the way to solve that
- 0:29problem is by creating your own model.
- 0:31I'll explain what it is and how to use
- 0:33it to make the AI tone disappear, very
- 0:35simply. I've prepared this so you can
- 0:37try it yourself after the video, so
- 0:39please watch until the end. Now, let's
- 0:41compare it with existing models; I put
- 0:43a lot of thought into making this
- 0:45comparison very fair. So, I ran it both
- 0:48ways. I covered a song generated with
- 0:50the latest version, 5.5, with my custom
- 0:53model, and covered a song made with my
- 0:55custom model using version 5.5. Before
- 0:58I play it, I'll let you know what you
- 0:59should focus on listening for. The AI
- 1:01tone doesn't come from the melody or
- 1:03chord progressions. Ultimately, it
- 1:04comes down to the difference in sound
- 1:05quality. Sound quality is directly
- 1:07related to these aspects. First is
- 1:09resolution: the clarity of the highs,
- 1:10the snare, and the drums, as well as
- 1:12those unnecessary noises in the high
- 1:14frequencies. And second is the image:
- 1:16it doesn't spread out, and the song
- 1:17sounds very stuffy. These are the true
- 1:19culprits behind that AI tinny sound.
- 1:21First, let's pass a song generated with
- 1:24version 5.5 through my custom model.
- 1:44Now, you can hear it clearly, right?
- 1:46The muddied details come back to life,
- 1:48and the instrument separation improves,
- 1:49making it sound much sharper. The
- 1:51stereo width increases, and the sound
- 1:53that was clustered in the center
- 1:54spreads out to both sides, sounding
- 1:56refreshing. If you listen closely,
- 1:58you'll notice the bass, drums, and lead
- 2:00instruments are organized, and the kick
- 2:01really pops. Songs made with Suno 5.5
- 2:03have clustered sound, and the
- 2:05instrument details are blurred. You
- 2:07might not notice it just listening to
- 2:08it, but for someone like me who makes
- 2:10music, it's immediately obvious. And
- 2:11when you compare the sound quality
- 2:13back-to-back like this, even the
- 2:15average person can easily tell the
- 2:16difference. That's why it feels
- 2:19well-made, yet somehow lacking. In the
- 2:21fields of music and composition, this
- 2:23detail is life, so it's a very
- 2:25important aspect. AI music is no
- 2:27exception. This time, let's do the
- 2:29opposite and pass a song made with the
- 2:31custom model through version 5.5. Let's
- 2:33listen to it right away. So, what do
- 2:53you think? The AI tone starts to creep
- 2:54back in, doesn't it? The resolution
- 2:56drops, the image narrows, and those
- 2:58characteristic AI noises in the high
- 3:00frequencies return. That fizzing,
- 3:01crackling sound. The aspects that just
- 3:03improved have moved in the exact
- 3:04opposite direction. After watching this
- 3:06far, you must be curious. So, how do
- 3:08you actually make this model? I'll show
- 3:10you right now. Well, to explain custom
- 3:12models simply, you're letting Suno hear
- 3:14your music. And then you train it on
- 3:16that. Instead of writing "make a sound
- 3:18like this" in text, you're uploading
- 3:20actual music files for it to hear.
- 3:22You're basically making your own
- 3:23version, like a version 5 or 5.5. Then,
- 3:25a dedicated model that reflects that
- 3:27sound is created. This is exactly the
- 3:29secret behind the sound difference we
- 3:31just compared. Making one is really
- 3:32easy. In the model dropdown, select '
- 3:34Create Custom Model,' set a name,
- 3:36upload music, and hit create—you're
- 3:38done. It finishes in a few minutes and
- 3:39appears in your list right away. But
- 3:41there is something important here. Does
- 3:42grabbing any six songs and uploading
- 3:44them make it all this good? Of course
- 3:46not. This is where the real quality of
- 3:48a custom model is determined. The
- 3:50reason the results improve is simple. A
- 3:52model only gets as good as the quality
- 3:53of the music it learns from. The reason
- 3:55AIT is so good is because I, as a
- 3:57composer, trained it on properly
- 3:58produced music. Conversely, if you put
- 4:00in six AI-generated tracks, you'll just
- 4:02get a hollow model. So, there are
- 4:04conditions the source material must
- 4:06meet. Consistent sound, uniform quality
- 4:08, and naturally, it should be good.
- 4:10Things like that. You just need to
- 4:11provide at least six tracks that meet
- 4:13these conditions. But Suno checks the
- 4:15copyright of uploaded music. Songs by
- 4:18favorite singers, famous BGM, and even
- 4:20free BGM are risky to use as training
- 4:22data. Most free music licenses only
- 4:25allow for use as video background music
- 4:27and do not account for AI training use.
- 4:30So, to sum it up, you need at least six
- 4:32tracks with excellent sound quality,
- 4:34consistent tone, and 100%clean
- 4:35copyright status, and those watching
- 4:37this far might be thinking: Okay, I
- 4:39know the feature is great, but where
- 4:41would an average person find those six
- 4:43tracks? You're right. So, I've prepared
- 4:46something so you can test it
- 4:47immediately after watching this video.
- 4:49I will share six original Pop R&B
- 4:51tracks that I produced myself. The
- 4:53quality is high, so you can upload them
- 4:55as they are. Download them, create a
- 4:57custom model, and try generating a song
- 4:59in the same style. You will feel that
- 5:01change I showed you earlier right away.
- 5:03Since it's for testing, there are
- 5:04license restrictions, and details are
- 5:06included in the file. The link is in
- 5:07the pinned comment. And one more thing:
- 5:09it's best to include as many
- 5:11high-quality tracks as possible. Six
- 5:13tracks is just the minimum requirement.
- 5:15As the number of tracks increases, the
- 5:16model becomes much more stable and the
- 5:18variation in results decreases. So,
- 5:20I’ve prepared a separate full set of
- 5:2119 tracks adjusted to the same tone.
- 5:24They are original wave files and come
- 5:25with a license that even allows
- 5:26commercial use. If you’re planning to
- 5:28monetize the models you create, you
- 5:29should check these out. I’ll leave
- 5:31this information in the pinned comment
- 5:32as well. Alright, I’ll continue
- 5:34sharing plenty of AI music tips that
- 5:35only a composer would know, so make
- 5:37sure to subscribe so you don't miss my
- 5:39videos. Please try generating them with
- 5:40the free version, and if they're good,
- 5:42I’d appreciate it if you left a
- 5:43comment. Thank you for watching until
- 5:45the end today. is.
About this transcript
This page contains the full transcript of 작곡가가 Suno 음질 3배 올리는 비밀 공식 (학습 음원 무료 배포) by 뮤잇 muitmusic, generated from the public captions YouTube serves with the video. The transcript has 1,071 words across 164 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.