SoundBoost: Prompt-Based AI Mastering Studio Launches New V4
- Ezra Sandzer-Bell
- 6 hours ago
- 6 min read
SoundBoost has reportedly crossed 180,000 musicians and creators, with more than 540,000 masterings completed and counting.
The AI mastering platform started out as a prompt-guided online mastering service, but the past few updates have pushed it somewhere more interesting. It's becoming an AI-powered studio for finishing, separating, understanding and practicing music.
We went through the latest release notes and spent time with the new features to make sure our reporting was accurate. There's a lot to cover here, so I've split it into two halves: the mastering side and the new stems and practice tools.
I spoke with their founder, Berkan Cesur. who explained to me that rebuilding the DSP has opened up new possibilities for SoundBoost. They can do real-time, local AI mastering and will have a separate VST for the same mastering engine. This will bring SoundBoost’s prompt-to-mastering workflow directly into DAWs, while keeping the processing consistent with the web and mobile versions.
Mastering Engine 4.0 gives you far deeper control than the earlier prompt-only workflow, the platform now runs on iOS and Android with one subscription across all devices, and a large chunk of the stem separation and practice experience is completely free. According to a recent press release, SoundBoost is currently serving 10,000 free stem splits every day.
Table of Contents
Mastering Engine 4.0: Text prompts + parameters
SoundBoost built its reputation on prompt-to-mastering. You upload a mix, type something like "warm, punchy, wide stereo field" and the engine assembles a signal chain to match. That workflow is still the core of the product, and it's still the fastest way I know to get from a bounced mix to a streaming-ready master.
What changed in Mastering Engine 4.0 is the depth underneath the prompt. Natural language input now resolves down to genuinely technical instructions.
You can ask for a boost at 15kHz and get exactly that, request a specific compressor style, or lean on improved reference mastering to chase the loudness curve and tonal balance of a track you love. Paste a Spotify link and the engine matches against it.
The part that surprised me most is stem-aware mastering. You can now remove or rebalance individual stems with a text prompt during the audio mastering process.
Tell it "remove drums and bring vocals more forward" and it just does that, no separate stem export and re-mix session required. For anyone who has ever received a stereo bounce with a mix problem baked in, this collapses what used to be a back-and-forth with the original producer into a single sentence.
Realtime Mode: talk to your mastering engineer
Grab your mic, open Realtime Mode, and describe out loud how you want your master to sound. The engine interprets your voice the same way it interprets a typed prompt, and your master is ready in seconds.
I was skeptical that voice input would add anything over typing, but it changes the feel of the session. Describing sound verbally is closer to how you'd actually brief a human engineer, and it keeps your hands free while you're A/B-ing against the original mix.
Working with AI Mastering Engineer Personas
I'm not a fan of what Suno has become in 2026, but the video above does a good job covering how Soundboost works in practice. You can use it for any kind of music, preferably human-made (and AI assisted when it saves you time without removing your creative agency).
Rather than relying on a single generic mastering approach, you pick the engineer that fits the record. Sophia leans into smoothness and clarity, Marcus is unapologetically a bass music guy, and the rest of the roster covers different stylistic territory. In practice this works like choosing which studio you send your record to, except the turnaround is about a minute.
The other workflow change in this release is bulk mastering. Until recently, SoundBoost handled one track at a time. Now every master generates a seed number, and you can copy that seed and apply it to any other track to transfer the same mastering character across songs.
If you're finishing an album or EP and want a consistent feel from track one to track ten, you master the single, copy the seed, and batch the rest. This was the single most requested feature from artists working on full projects, and it's easy to see why.
SoundBoost on iOS and Android
SoundBoost is now available on both iOS and Android, bringing the mastering and stem tools to mobile. One subscription covers web, iPhone, iPad and Android, and your projects sync across devices.
Mobile mastering used to feel like a gimmick to me, but the use case is real: reference your master in the car, on earbuds, on a bluetooth speaker, and push revisions from wherever you are. Combined with Realtime Mode, you can literally talk a revision into existence from your phone.
One note on the apps: the stem splitter has its own dedicated iOS app, AI Vocal Remover & Stem Splitter, which is separate from the base SoundBoost app. An Android version of the stem splitter app is coming soon.
Stems & Studio: guitar and piano separation
The free stem splitter has been quietly turning into its own product. Vocals, drums, bass and other standard stems have been supported for a while, and the free vocal remover remains one of the cleanest ways to pull an acapella from a stereo file. The new addition is guitar and piano separation.
Guitar and piano are historically the hardest sources to isolate because they occupy the same midrange territory as everything else in a mix. Having them as dedicated stems is perfect for instrumental practice and backing tracks. Pull the guitar out of a song you're learning, play along with everything else, then bring the original part back in to check yourself against it.
Lyrics Mode, Chords Mode and the adaptive metronome
This is where SoundBoost stops looking like a mastering service and starts looking like a practice studio.
Lyrics Mode is a built-in karaoke mode aimed at musicians learning or practicing songs. SoundBoost extracts the lyrics from a track and displays them in sync while it plays. If the key of the track doesn't sit right with your voice, you can change the pitch without changing the tempo and sing it where it's comfortable.
Chords Mode detects the chords in a song and turns it into an interactive practice experience for guitar, piano and ukulele. Instead of hunting down a chord chart of questionable accuracy, you get the progression mapped against the actual recording.
Looping and remixing tools round out the practice workflow. You can independently change pitch and tempo, loop sections, isolate stems, adjust volume and panning, and drill specific parts of a song. When you've got a configuration you like, you can bounce a mix from the currently audible stems, with solo and mute states respected. That means custom backing tracks in a couple of clicks.
The adaptive metronome deserves special mention. Most metronomes assume a song was recorded to a fixed grid, which rules out a huge portion of recorded music, especially anything tracked live before click tracks were standard. SoundBoost's metronome follows the track itself, drifting with the performance, which makes it genuinely useful for practice and live playing along with real recordings.
What's free and where to start
A lot of the stem separation, lyrics and practice experience is available at no cost, including the free stem splitter and vocal remover, which is now handling 10,000 free splits every day. If you want to hear the mastering engine before committing to anything, there's a free AI mastering tier online as well, no signup required for a preview.
The overall direction is clear. SoundBoost began as an online mastering tool and it's still one of the strongest audio mastering options in that category, but the free stems, practice modes and mobile apps point toward something broader: one place to finish a record, take it apart, understand it and learn to play it.
You can explore everything at soundboost.ai.
