ñü.ŵævẽš
every instrument, where it belongs.
spacetime bubble™
Concert hall. Strings, winds and piano, written as a film score.
a 2008 demo
rebuilt in space.
First Attempt was made in GarageBand in 2008 and all that survived is a small stereo AAC file. nu.waves picked out each instrument, split it from the others and set them in a concert hall. Put on headphones, press play, switch to the original to hear where it started, then drag the instruments anywhere around you.
- Original 2008
- AAC-LC · stereo · 44.1 kHz · 191 kbit/s · 9.6 MB · 6:40
- ñü.ŵævẽš
- 5 instruments in 3D · 48 kHz · 174 kbit/s · 9.1 MB · 6:40
- winds & strings 48 · drums 45 · piano 39 · strings 21 · bass 20 kbit/s
- Opus 5.1
- 6 channels · 48 kHz · 258 kbit/s · 13.0 MB
- Lossless FLAC of the original
- stereo · 44.1 kHz · 16-bit · bit-exact · 28.6 MB
The nu.waves file on this page is 9.1 MB, 9% smaller than the 2008 stereo file and 33% smaller than the Opus 5.1 file. Those are file sizes. Listening quality is nu.fidelity, scored by ear. nu.waves is the one you can move, to any angle, distance and height. 5.1 fixes six speakers on one flat ring, with the instruments mixed together inside them.
The source is a 6:40 stereo AAC from 2008. Demucs, an open-source separation model, ran on a Mac and found five parts in it: drums, piano, strings (violin, viola and cello, many takes of each), winds with strings, and bass. It stands in for the on-device model. Nothing was re-recorded and nothing was added: each instrument became its own sound, set in the room and stored as a nu.waves object.
Five instruments you can move anywhere take 9.1 MB, less than the 9.6 MB original stereo file and 30% less than the fixed 5.1 mix this page used to play. Each instrument is one sound at one point, so it needs one channel, not six.
Any headphones. This page renders the room for your two ears in the browser, with no head tracking. In the khėnsu.ñü app, nu.waves hands the same objects to Apple’s spatial audio on AirPods, so they stay put when you turn your head.
one sound, then where it sits.
nu.waves stores each instrument once. Where it sits, how high, how far, and the room around it, is a list of numbers beside the sound. Moving it, or scheduling a move, does not encode the audio again and does not add channels.
Opus 5.1 on this page is the same song folded into six fixed speakers: 258 kbit/s, 13.0 MB, measured. Opus at 48 kbit/s is a stereo rate used when a song should still sound whole. A 5.1 mix packed into that same 48 kbit/s has about 8 kbit/s for each speaker. nu.waves on this page is 174 kbit/s, measured, because each instrument here is still an Opus stem. The listening score is nu.fidelity, scored by ear, and it is separate from that rate.
The aim is a much lower packing rate at the same listening score. That aim is an estimate until the listening test is run. What is already true: a path through the room is lossless next to the samples. An artist can schedule the move. A listener can schedule their own. The person stays put, the sound moves, and the samples do not change, so the path costs a few numbers, not another mix.
in the ear.
nu.ears are in-ear. No camera. Voice access stays on the device. A call goes out through two relays run by different operators, the same rule as every khėnsu.ñü call, and those relays carry voice and audio.
They play a nu.waves object in each ear, so height and distance are rendered there instead of folded into a stereo mix first. The spec is louder than AirPods Pro 2, and a longer charge than its 6 hours of listening with noise control on.
AirPods Pro 2 listening time is Apple’s published figure. Loudness and hours for nu.ears are the product spec; measured figures replace them when a pair is tested. AirPods is a trademark of Apple Inc.
in the room.
Every sound has its own height, above or below your ears as well as all the way round. Pick a track and the player follows it. Stair Creek sits beside you. Wilhelm Scream is the 1951 yell from Wikimedia Commons (CC0). The chakra tones run from the root, below you, up to the crown: Muladhara 194.18 Hz, Svadhisthana 210.42 Hz, Manipura 126.22 Hz, Anahata 136.10 Hz, Vishuddha 141.27 Hz, Ajna 221.23 Hz just in front of your forehead, Sahasrara 172.06 Hz. Each tone holds, fades, then ends, or hands off to the next with a crossfade. Auto-cycle walks them from the root to the crown. Clone and remove sit on the sound you have selected in the spacetime bubble.
Drag a sound to move it anywhere in the spacetime bubble, up to 4 m away, above or below you; drag the bubble itself to look around it. The height slider sets the selected sound’s height. With a keyboard, use the arrow keys, and Page Up and Page Down for height. Each sound sends its own waves across the bubble as it plays.
Gain 0 dB keeps nu.waves as loud as the original 2008 file, measured as you listen. Reverb is the room. Echo duration is how long it lasts, echo delay is the gap between repeats, and echo bounces is how many you hear. Clean takes the room off. These five controls are nu.eq.
ñü.ŵævẽš · compared
smaller than the original. and it moves.
The same 6:40 song, four ways. nu.waves carries every instrument as its own sound in 3D, in less space than the 2008 stereo file, and it’s the only one here you can rearrange while it plays.
- ñü.ŵævẽš9.1 MB
- Original 20089.6 MB
- Opus 5.113.0 MB
- Lossless FLAC28.6 MB
- ñü.ŵævẽš174
- Original 2008191
- Opus 5.1258
- Lossless FLAC571
- Winds & strings 48
- Drums 45
- Piano 39
- Strings 21
- Bass 20
| What you can do | ñü.ŵævẽš | StereoAAC, FLAC | Opus 5.1 | Dolby Atmos |
|---|---|---|---|---|
| Every instrument its own sound | ● | – | – | ◐ |
| Move any instrument while it plays | ● | – | – | – |
| Sounds above and below you | ● | – | – | ● |
| Distance: near or far, up to 4 m | ● | – | – | ◐ |
| Pick the room as you listen | ● | – | – | – |
| ñü.ÊQ: gain, reverb, echo duration, delay and bounces | ● | – | – | – |
| Plays on any headphones, in any browser | ● | ● | ◐ | ◐ |
| Switch to the untouched original at any moment | ● | ● | – | – |
| Smaller than the original stereo file | ● | – | – | – |
| Duplicate a track and shift it in space and timecoming | ● | – | – | – |
| Artist-locked sound per songcoming | ● | – | – | – |
Sizes and rates measured on the files this page plays, 12 October 2026. FLAC is the original losslessly compressed, 16-bit, and bit-exact. Each nu.waves instrument on this page is Opus at 48 kHz inside the format. The charts count bytes. nu.fidelity is the listening score, scored by ear, and it is separate from kbit/s. Dolby Atmos places sounds in height too; its mix is set by the engineer. Dolby and Dolby Atmos are trademarks of Dolby Laboratories; no affiliation.
ñü.ÊQ · akashic
all audio, placed in the room it was meant for.
What you just heard, done for every song, talk and film without a spatial mix of its own. nu.waves listens, picks out every instrument and voice, works out the room, and places each one there, on your own iPhone, iPad or Mac. Nothing is sent anywhere. For Akashic members.
- Concert hallStrings and choir across the stage, two seconds of hall around you
- StadiumThe band far ahead and high, the crowd and the echo all the way round
- FestivalKick and bass dead centre, synths wrapping around and above you
- ClubDrums a few steps away, keys and guitar either side
- Private roomA voice and a guitar an arm’s length away
- Book readingOne voice across a quiet table, the room barely there
- InterviewTwo voices facing each other, the host on one side, the guest on the other
- FilmDialogue on the screen, music around you, effects where they happen
Sound recognition on the Neural Engine hears what is playing: which instruments, which voices, music or speech, and how big the room was. That decides the venue, and it tells you why. Never Private Cloud Compute.
Each instrument and voice becomes its own object you can move, like the spacetime bubble above. Bass stays close and centred, and the air above 6 kHz opens up and rises, the way your ears expect.
The goal: the parts add back up to the original, so folding it to stereo gives you the artist’s master. Today’s split doesn’t reach that yet, so one tap always plays the original itself, and artists can switch nu.eq off for their work.
nu.waves files use the .nu container: the header saying what is inside is encrypted along with the sound, so a file looks like random bytes to anything except the app. Video works the same way.
sounds
sounds.music and healing, straight from the artist.
A new tab in khėnsu.ñü, next to Chats and Feed. Artists upload their own music. Every play pays the artist 1¢ with no cut, fans buy albums and tickets straight from the artist, and it all plays privately.
Artists you back
Fresh uploads
Try it: switch to Healing, or tap anything to play it.
- Upload. Sell or stream.Put out a track, an album or a live set yourself. Sell it, stream it, or both.
- Paid for every play.Every play pays the artist 1¢, $10 per 1,000, and khėnsu.ñü takes no cut. You set the price of what you sell, such as albums, downloads and tickets, with no platform cut.
- Your rights stay yours.Masters, publishing and copyright stay with you. khėnsu.ñü takes no licence and no share.
- A healing library.Solfeggio frequencies, binaural beats, rain, ocean, singing bowls and sleep sounds, from the people who record them.
- Private by design.Everything plays over the stealth network. Nothing is processed on servers, so nobody builds a profile of what you play.
- Offline, too.Download anything you’ve bought, and play it with no connection at all.
Artists set their own prices for albums, downloads and tickets; every play pays the artist 1¢, with no cut. Paid content bought inside the iPhone app follows Apple’s purchase rules; tips go 100% to the artist.
Every play pays the artist 1¢ with no cut, and what you buy is yours to keep offline.See what creators keep