You have a Suno or Udio song you want on Spotify, Apple Music and YouTube, and you want it to hold up next to everything else in a playlist. Mastering for streaming is mostly about two numbers, integrated loudness and true peak, plus one habit: measure what you have before you change it. This guide gives the figures each platform publishes, exactly as they publish them, the order to work in and a checklist you can run on any track.
We make an automatic mastering service and audio clean-up tools, and we say where they fit and where they do not. Every platform figure below comes from the platform’s own page, listed under Sources.
The short answer
Measure the render, clean any artifacts, then master to the platform’s published guidance. For Spotify that is -14 LUFS integrated with true peak below -1 dBTP, or below -2 dBTP if the master is louder than -14 LUFS. Apple publishes no LUFS target but asks for at least 1 dB of headroom and 24-bit sources; the YouTube help page we read gives no loudness figure. Check the master through a lossy encode before you deliver.
What each platform publishes
Streaming services do not process your file to a new loudness. They measure it and change the playback level. Spotify says so directly: loudness normalization is applied during playback, and it does not process or change your audio before playback.
| Platform | Playback level | Mastering guidance it publishes |
|---|---|---|
| Spotify | Adjusts tracks to -14 LUFS (ITU 1770). Premium listeners can pick Loud (-11), Normal (-14) or Quiet (-19 LUFS). Albums are normalized as a whole; shuffled or playlist tracks individually. | -14 LUFS integrated; true peak below -1 dBTP; below -2 dBTP for masters louder than -14 LUFS. The same guidance applies to lossless playback. |
| Apple Music | Sound Check stores a track’s loudness in metadata and raises or lowers playback volume; it can also set volume per album. No numeric target published. | Leave at least 1 dB of headroom to avoid clipping when playback oversamples; send 24-bit sources at the highest native sample rate; do not upsample. |
| YouTube | Stable volume is on by default for videos and continuously adjusts volume; it is turned off for YouTube Music and official music videos. | No loudness figure on the help pages we read. |
Three details from Spotify’s page change how you should think about level. A master louder than -14 LUFS is turned down, and Spotify says no additional distortion occurs from that. A quieter master is turned up only as far as its true peak allows, leaving 1 dB of headroom for lossy encodings: its own example is a -20 LUFS track with a -5 dBFS true peak, which is lifted only to -16 LUFS. And the web player and third-party devices such as TVs and speakers do not normalize at all, so some listeners hear your file at its delivered level.
Apple’s brief makes the same point in plain words: with Sound Check and similar technologies, songs mastered loud are played back at a lower volume, which can make them sound weaker. It asks you to master for your intended sound regardless of playback volume.
Why an AI render needs measuring first
Generators do not publish a loudness target for their output that we could find; Suno’s download help page does not mention one. We have not measured a set of Suno or Udio renders ourselves, so we will not tell you they all arrive loud. What you can do is check yours in a minute:
- If the integrated loudness is already well above -14 LUFS and true peaks sit near 0 dBFS, the render has probably been limited already. Adding a second limiter on top stacks more gain reduction for no gain on Spotify, which turns it back down.
- Spotify’s page notes that louder tracks are more susceptible to extra distortion when encoded for streaming. A render louder than -14 LUFS is the case its -2 dBTP advice covers.
- If it is quiet and dynamic, you have room to work, and Spotify will only lift it as far as its peaks allow.
Start from the best file you can get. Suno’s help center lists MP3 downloads on all plans and WAV downloads for Pro and Premier subscribers; mobile downloads default to MP3. Converting an MP3 to WAV does not restore what the MP3 discarded (see WAV vs MP3 for AI music). For Udio, its help center article on the Universal Music Group partnership (which it says began October 29, 2025; article dated February 17, 2026) says downloading of audio, video and stems has been disabled; check the current status before you plan around a Udio file.
A free way to measure: ffmpeg’s ebur128 filter reports integrated loudness and loudness range, and with peak=true it measures true peak on an oversampled copy of the signal. For example, ffmpeg -nostats -i song.wav -af ebur128=peak=true -f null - prints a summary at the end. Most DAWs also ship or support a loudness meter that does the same job.
Clean artifacts before you master
Mastering raises the level and narrows the dynamics of everything in the mix, including the metallic shimmer or watery high end some AI renders carry. Clean first, then master the cleaned file. Our AI music restoration order guide places structural cleanup before EQ, de-essing and compression, and ends with a matched-loudness check and a lossless master; mastering belongs after all of it.
If your render has that residue, our de-artifact plug-in (VST3, AU and CLAP, 14-day trial) and de-artifact Cloud in the dashboard ($0.25 per minute of audio from your Intrect credit) are made for it. Strong settings can also remove musical detail, so listen to what is taken out. Spectral repair in a general restoration editor is another route. Cleanup changes how a track sounds; it is not a way to change how a platform or detector classifies it.
A mastering checklist for streaming
- Measure the source. Note integrated loudness, true peak and loudness range of the render before you touch it.
- Clean first. Fix artifacts, clicks and bleed on the mix or stems, and keep a lossless copy of the cleaned file.
- Leave headroom. Keep true peak below -1 dBTP, and below -2 dBTP if you master louder than -14 LUFS (Spotify). Apple asks for at least 1 dB of headroom.
- Choose loudness on purpose. -14 LUFS integrated matches Spotify’s guidance. Mastering louder is a creative choice; Spotify’s normalization and Apple’s Sound Check will turn it down for listeners who use them.
- Preview the codec. Spotify names Ogg/Vorbis and AAC as its lossy formats. Apple’s mastering tools afclip and AURoundTrip are included in current macOS, and its free AU Lab app can host the AURoundTripAAC plug-in and help detect peaks and clipping. Listen for new distortion after the encode.
- Compare with references at matched loudness. Turn your reference tracks and your master to the same loudness before you judge tone and punch, the way a normalizing player will.
- Deliver one high-quality stereo master. Spotify says one per track is all it needs. Apple asks for 24-bit at the native sample rate, without upsampling or padding a 16-bit file.
- Read the distribution rules. For what Spotify, Deezer and YouTube say about AI-generated music, labels and disclosure, see does Spotify, Deezer or YouTube flag AI music?
Where Intrect Mastering fits, and where it does not
Intrect Mastering is automatic mastering in the dashboard. You upload a mix (WAV, FLAC, MP3, AIFF, M4A or OGG, up to 768 MB) and get a free 60-second before/after of its loudest section, usually within one to two minutes; previews are free, a few per day per account. If you keep it, the full master costs a flat $3.00 per track, your first full master is free (one per account), and an order of several tracks gets 10% off at 2–4 tracks, 20% at 5–9 and 25% at 10–20, up to 20 tracks per order. It is paid from your Intrect credit, and a failed job is not charged. Each master comes in two versions: a Streaming version, whose level is set per song and keeps more dynamics, and a louder CD version for CD and downloads. The dashboard opens on the Streaming version, shows the measured loudness of each version and groups the downloads by version, each as WAV 44.1 kHz 16-bit, FLAC and MP3 320 kbps.
Read its limits against the table above:
- The Streaming version is not pinned to -14 LUFS. Its level is set per song, so it can still come out louder than Spotify’s figure, and Spotify will then turn it down. Read the loudness the dashboard reports for it before you deliver. The CD version is louder again and is meant for CD and downloads, not for a normalizing player.
- The dashboard does not show true peak. Measure the downloaded file’s true peak yourself, for example with the ffmpeg command above, and compare it with Spotify’s -1 and -2 dBTP guidance.
- The downloads are 16-bit at 44.1 kHz. Apple asks for 24-bit sources at the native rate, which these files do not meet.
If you want a master that lands exactly on -14 LUFS with true peak below -1 dBTP, or a 24-bit delivery for Apple, master in your DAW with a true-peak limiter and a loudness meter, or use a mastering engineer or service that delivers 24-bit files. The free preview is a cheap way to hear whether an automatic master suits the song before you decide.
Clean first, then hear a master
Check the restoration order for your track, then listen to a free 60-second mastering preview of your own mix.
Related guides
- AI music restoration: what to run first
- Does Spotify, Deezer or YouTube flag AI music?
- WAV vs MP3 for AI music
- Suno artifacts
- Udio artifacts
- Intrect Mastering
- Pricing
Sources
- Spotify for Artists, Loudness normalization on Spotify, support.spotify.com/us/artists/article/loudness-normalization, read October 25, 2026.
- Apple, Apple Digital Masters technology brief (PDF), apple.com/apple-music/apple-digital-masters/docs/apple-digital-masters.pdf, read October 25, 2026.
- Apple, Apple Digital Masters tools page, apple.com/apple-music/apple-digital-masters, read October 25, 2026.
- YouTube Help, Control video volume on your device, support.google.com/youtube/answer/14106294, read October 25, 2026.
- Suno Help, What types of files can I download?, help.suno.com/en/articles/13926081, read October 25, 2026.
- Udio Help, Changes associated with the Universal Music Group partnership, help.udio.com/en/articles/12683565, read October 25, 2026.
- FFmpeg filters documentation, ebur128, ffmpeg.org/ffmpeg-filters.html#ebur128, read October 25, 2026.
- Our Intrect Mastering and pricing pages.
FAQ
What loudness should an AI song be mastered to for Spotify?
Spotify's loudness page recommends a master at -14 LUFS integrated with true peak below -1 dBTP, and below -2 dBTP if the master is louder than -14 LUFS. Spotify turns louder tracks down to -14 LUFS at playback, so a louder master is not louder for listeners who keep normalization on.
Does Apple Music publish a LUFS target?
Apple's Apple Digital Masters technology brief does not give a LUFS target. It describes Sound Check, which raises or lowers playback volume from a stored loudness value, asks for 24-bit sources at the highest native sample rate and recommends leaving at least 1 dB of headroom to avoid clipping on playback.
Should I clean artifacts before or after mastering an AI song?
Before. Mastering raises the level and compresses the dynamics of everything in the mix, including metallic or watery residue, so clean the residue first and master the cleaned file. Our restoration-order guide puts normal processing after the structural cleanup.