Mix finished? Upload it — Magic Master masters your track online: LUFS targets for Spotify (−14), Apple Music (−16) and YouTube (−14), True Peak ≤ −1 dBTP, 24 presets. Download a streaming-ready master. Built for AI music (Suno, Udio) too.
Studio-grade AI mastering from any country: the same LUFS, True Peak and streaming readiness. Buy tokens in the wallet with a card (YooKassa); a card issued outside Russia may be declined — then write to support@magicmaster.pro. Tokens are credited to your account automatically.
Our first poll, "What do we build next?", picked two options, and both are already built. Plus this week's fixes since November 9. • Gentle …
Our first poll, "What do we build next?", picked two options, and both are already built. Plus this week's fixes since November 9. • Gentle vocal pitch correction is live now (magicmaster.pro/en/app/vocal-pitch). The voice is pulled toward the nearest note or the steps of the track's key; vibrato and slides stay, with no robot sound. Presets "Gentle" and "Noticeable". The first 30 seconds are a free preview; the full version needs a sign-in: 1 token per run, for a recording up to 5 minutes. • Backing vocals and doubles come next: a double, a third up or down, a fifth up, in the key of the track. • How the votes went: pitch correction 12, backing vocals 6, EQ in plain words 4, one-sound album 2, synced lyrics 1, iPhone app 0 (as of November 15, 12:30). Thank you to everyone who voted. The next poll opens in the app. • Track analysis progress is now in one place, in the AI advice block: the upload with a percentage, the analysis and the advice. • A more compact upload box on phones. • The Master button no longer stays silent: if there is a question before the run, the app scrolls to it and tells you what to pick. • In the bot, linking an account with digits after /link works again, and a big WAV comes as a personal link for 7 days. Poll results: magicmaster.pro/blog/poll-1-results-pitch-correction-backing-vocals This week: magicmaster.pro/blog/magic-master-weekly-2026-11-15
Version 3.20 came out on November 9. It brings a new look of the app and three tools for long recordings and speech. All of it is in the app…
Version 3.20 came out on November 9. It brings a new look of the app and three tools for long recordings and speech. All of it is in the app now. • A new look, Tiles and Pro. Tools and settings are compact tiles, and fine tuning folds into the Pro panel. Four themes: Classic, Night, Soft and Contrast, plus System. The progress ring, track passport, before/after player, variants and formats are all there. • Long tracks and audiobooks up to 2 hours in one file (magicmaster.pro/app/master): audiobooks, podcasts, lectures, DJ sets. Up to 10 minutes costs 1 token, and every started 15 minutes adds 1 more. You see the price before the run and can close the tab: we email you when it is ready. If processing fails, the tokens come back. • Audio to text (magicmaster.pro/app/transcribe): text with timecodes, TXT, DOCX, SRT and VTT subtitles, a summary, topics and action items. Russian speech. 30 free minutes a day when signed in, 5 without an account, then 1 token per started 30 minutes. • Refine with AI advice. On the result screen the AI suggests settings for a new master: style, loudness, bass and treble. You see what will change and what it costs before the run. • Smart speech cleanup in the audio cutter (magicmaster.pro/app/cutter): "um" and "uh", breaths, mouth clicks and level jumps are marked on the waveform. 5 free minutes a day, then 1 token per started 10 minutes. • AI trace cleanup and mastering in one run, with your own mastering settings and no second upload. More: magicmaster.pro/blog/magic-master-3-20-whats-new Tell us what works and what does not: your feedback shapes the next release.
Magic Master now masters long recordings: audiobooks, podcasts, lectures, DJ sets and live shows up to 2 hours in one file. It is part of ma…
Magic Master now masters long recordings: audiobooks, podcasts, lectures, DJ sets and live shows up to 2 hours in one file. It is part of mastering in the app: magicmaster.pro/app/master How it works • Upload a file longer than 10 minutes and the app shows the price and the processing time before you start. • Pick the type: Voice, Podcast, Audiobook or Music. Audiobook aims at −19 LUFS with peaks no higher than −3 dBTP (the ACX guide). • Loudness and tone are set from the whole file, then the track is processed in one-minute parts and joined with crossfades. There are no loudness jumps at the joins. • Processing runs on our server. You can close the tab: we will email you and send a Telegram message if it is linked. The result stays in the app for 72 hours. • If processing fails or you cancel it, the tokens come back. Price • Up to 10 minutes: 1 token, like regular mastering. • Every started 15 minutes after that: +1 token. Output: WAV 24-bit, FLAC or MP3. Step by step: magicmaster.pro/blog/long-audiobook-mastering-one-file
Version 3.19 came out on November 7. The main idea: before you run mastering you can see what will be done, and after it you can see what ch…
Version 3.19 came out on November 7. The main idea: before you run mastering you can see what will be done, and after it you can see what changed. All of it is in the app now. Mastering • AI advice. In the track studio, right after upload: a style, a loudness and the reason in one sentence, with a button that starts mastering from that advice. • A new mastering screen in every entry: "Auto will do: style and loudness" up front. Four sound variants appear on their own after analysis with matched loudness, and a tap picks the style (signed-in users get 30 previews a day). "Soften harshness" and "Louder and denser". Format and sample rate 44.1 / 48 / 96 kHz. On a desktop, two columns and one "Master" button always in view. How to choose: magicmaster.pro/blog/choose-mastering-sound-before-you-master • Processing you can follow: upload and processing shown separately, a time estimate, seven named steps, the track waveform and "Finalizing" on the last percent. After a site update the open tab offers to refresh itself. • The result starts with a summary: what was done, the style, loudness before and after, the format and the sample rate. Main button "Download WAV 16-bit 44.1 kHz". "Not right? Try differently": a Spotify or Apple Music version, louder or quieter, bass and treble. Charts "Loudness over time" and "Spectrum before/after", and one "Original / Master" player with matched loudness and a loop. • Sound. In the Universal and Trap styles there is more air and less dullness: the tonal correction switches on only when a defect is found. Checked on 159 tracks; loudness and peaks do not change. The share of tracks with an air cut of 1 dB or more fell from about half to about 10%. More: magicmaster.pro/blog/dull-master-universal-trap-check AI trace cleanup and vocal removal • AI trace cleanup (magicmaster.pro/app/aiclean). Before the run it says plainly that it weakens audible generator marks and hiss, and it does not promise that detectors will not find the track. If no marks are found, the token is returned and the track can be mastered at once. "Download cleaned WAV" and "Master the cleaned track" on top, an "Original / Cleaned" player. You can come back to the task within 24 hours after closing the tab. In the studio there is one run: "Remove AI trace and master". • Vocal and instrumental (magicmaster.pro/app/stems). "Download instrumental" and "Download vocal" on top, an "Original / Instrumental / Vocal" player, "Listen to 30 s" before processing ends, "Tell me when it is ready" by email. "Put a voice on this instrumental": the instrumental is already loaded. Tools in the app • Album (magicmaster.pro/app/album): several tracks, one loudness and sound, processing strength. • Stem mastering (magicmaster.pro/app/stems-master): 2–8 stems of one track. • Mastering calculator (magicmaster.pro/app/mastering-calculator) and loudness penalty (magicmaster.pro/app/streaming-penalty) with listening after the platform's normalization. • Restoration (magicmaster.pro/app/restore): cassette, vinyl, phone and interview sets and fine tuning. • A processing chain in "More settings": modules, strength, my presets and community presets. • "More about the track" before the run: spectrum, stereo and phase, how each platform will turn it down, reports. • AI clip (magicmaster.pro/app/clip_ai): 30 s, 60 s or the whole track, standard or premium quality. A clip can be made straight from a finished master without re-uploading. Convenience • All formats the server understands are accepted: wav, mp3, flac, m4a, ogg, aac, aiff, aif, webm, opus, amr, oga. • The app can be installed on a phone and a computer. • In your account: password change, API keys, history export to CSV, account deletion, "Invite a friend" and achievements. • "What's new" on the home page and compact "Recent tracks" with playback. • A finished master can be delivered to Telegram. • The weekly vote "What do we build next?" (magicmaster.pro/app/vote): signed-in users vote, one vote each, extended to November 15. The wishes form shows the hint "It can already do this". • The AI consultant sees the open track. Tell us what works and what does not: your feedback shapes the next release.
Loading...
How it works
Three steps to a finished master
Pro results with no engineering guesswork. The pipeline picks settings for your genre and target platform.
01
Upload your track
Drop WAV, MP3 or FLAC on the page. Free, no sign-up, files up to 160 MB; WAV up to 1000 MB after a token purchase.
02
Pick a genre
Universal, EDM, Hip-Hop, Classical, Podcast, Lo-fi, House, Dry Vocal. LUFS, EQ and compression match the style.
03
Download the master
Get your master (MP3 320 right away; WAV and FLAC after a free sign-up) with normalized level, dithering and controlled peaks.
What you get
Numbers you can check
The values come from the mastering settings themselves, not from promises.
Loudness per platform
Spotify — −14 LUFS
Apple Music — −16 LUFS
YouTube — −14 LUFS
VK Music — −14 LUFS
Yandex Music — −14 LUFS
Zvuk — −14 LUFS
The target is the playback level for each platform. Platforms already turn down over-loud tracks, so the master does not push louder.
True peak
True Peak no higher than −1 dBTP. The ceiling is the last step before export.
Modules included in every token mastering, described below.
Reference mastering
Reference Track
Upload a reference — we build and apply a corrective EQ curve via FFT spectral analysis. Your track matches the tonal balance you love.
Adaptive correction
Dynamic EQ — 8 bands
EQ moves only when needed — per-band level following. Often 3–5× more transparent than multiband compression. Tames mud and resonances without killing character.
Dynamics control
Transient Designer
Independent attack and sustain via dual envelope followers. Punch up drums or soften acoustic transients.
Noise cleanup
Spectral Denoiser
Frequency-domain Wiener filtering removes noise, hum and recording artifacts. Median-spectrum noise profile — stable on transients, preserves highs.
NY compression
Parallel Compression
Blend heavily compressed signal with the dry path. Keeps transients, adds density — classic studio technique for musical weight without squashing.
Before release
Streaming Loudness Preview
See how Spotify, YouTube, Apple Music, Tidal and broadcast would normalize your track — green/yellow/red hints.
Magic Master combines Reference Mastering + Dynamic EQ + Transient Designer + Streaming Preview in every token-based mastering. No other online mastering service offers all of these at once.
Suno, Udio and other AI generators leave inaudible high-frequency markers in tracks — a digital fingerprint that reveals machine origin. Magic Master analyzes and removes these markers while preserving sound quality.
HF Steganography
Suno embeds inaudible patterns above 16 kHz. We cut them with an 8th-order Butterworth filter — without losing musical content.
Block Artifacts
AI builds music in tokens — energy jumps appear at block boundaries. Crossfade smoothing removes these traces.
Generator fingerprint
The network's decoder leaves narrow spectral peaks at fixed frequencies. A 7-cent pitch shift — less than a tenth of a semitone — moves them off their usual spots; length stays the same.
A full mastering chain. Order and parameters are yours to tweak.
LUFS normalizationFree
Spotify −14, Apple −16, club −9, broadcast −24. True Peak ≤ −1 dBTP.
Studio EQFree
Industry studio curve. HP 40 Hz, LP 18 kHz. Linear or minimum phase.
4-band compressionFree
Multiband dynamics with soft knee and upward compression. Adjustable crossovers.
Spectrum analyzerFree
4096-point FFT, 64 log-spaced bands 20 Hz–20 kHz.
VectorscopeFree
Lissajous L vs R up to 6000 points. Stereo width and phase at a glance.
Harmonic exciterPRO
Tape/tube/warm/transistor/digital modes. 1×/2×/4× oversampling. Saturation without clipping.
Stereo widthPRO
4-band M/S imager. Stereoize delay. L/R correlation in the track analysis.
Reference matchingNEW
FFT spectral match to a reference → corrective EQ. Sounds like your favorite record.
Dynamic EQNEW
Eight bands follow level. Clearer than multiband comp. Tames mud and resonance.
Transient shaperNEW
Independent attack and sustain. Add punch or soften transients.
DenoiserNEW
Wiener denoising in the frequency domain — noise and hum without losing detail or air.
Parallel compressionNEW
NY-style blend of crushed and dry. Weight and vibe without squash — studio standard.
ReverbPRO
Five types: plate, room, hall, theater, cathedral. Separate M/S wet/dry.
TPDF + noise shapingFree
Dither for 16-bit export: TPDF, ns_e, ns_itu. Auto-blank.
Module chainPRO
Drag-and-drop order, per-module amount, JSON API configuration.
LUFS over timeFree
400 ms loudness blocks. Spot trouble areas before mastering.
Tokens come back
If a master fails, the token is returned to your balance automatically. You pay only for results you actually got.
AI assistant
Reads your track's metrics and explains its choice: \u201ccrest factor 6.4 dB \u2014 the track is over-compressed, use Already loud\u201d. Suggests a preset and a platform target.
Already loud
Restores dynamics to tracks crushed into a brickwall — from Suno, Udio or an over-aggressive mix. It does not compress harder; it opens the punch back up.
Presets for speech
Voice for narration and video (−16 LUFS) and Audiobook to the ACX standard (−19 LUFS): intelligible voice, rumble filter, no reverb.
A master in about a minute
A four-minute track on the standard preset takes about a minute.
REST API accessPRO
Full REST API with Swagger. Wire into your DAW, scripts, CI/CD.
Pricing
Clear pricing no surprises
1 token = 1 full PRO master from 35 . No subscriptions, no recurring fees.
Free
Free
Freeforever
Free account — 1 basic mastering per UTC day (UTC)
Magic Master is an online AI audio mastering service at magicmaster.pro: upload in the browser, LUFS targets for Spotify/Apple Music/YouTube, True Peak ≤ −1 dBTP, 24 presets; Telegram bot and REST API included.
Do I need an account for free mastering?
No. The free mode works without registration — upload a file and get the result. Limits: guest 1 per day (UTC), account 3 per month (see pricing).
How many free masterings do I get?
Guests without sign-in: 1 basic master per day (UTC). Free account: 1 basic mastering per UTC day. All formats after sign-in; PRO modules — with tokens. Remaining limits are shown in the app.
Which formats are supported?
Input: WAV, MP3, FLAC, M4A, OGG, AMR, AAC, AIFF and WebM. Export: guests get MP3 320 kbps; WAV 16-bit and FLAC after a free sign-up; WAV 24-bit, MP3 320 kbps and FLAC 24-bit with PRO token mastering. 24-bit is the master format for distribution. MP3 needs FFmpeg on the server.
What is LUFS and why normalize?
LUFS is the loudness standard for streaming. Spotify ~−14, Apple ~−16, clubs ~−9. Magic Master targets the level you choose.
Are my files stored on your servers?
Sources are processed in memory. The finished master is kept on the server: free account — 1 day, after token payment — 30 days. Guests keep history only in the browser (localStorage).
Can I edit the processing chain?
Yes, when you pay with tokens: drag-and-drop module order, per-module amount, linear-phase EQ, reverb type and M/S settings.
Is there an API for my workflow?
Yes. Full docs at /docs. REST API runs mastering, analysis and metrics. The API is available to all accounts.