The free, open-source Mac player for surround and hi-res music collections.
Vespertine plays your stereo and surround music from one library: SACD images, DVD-Video discs, DTS CDs, Dolby TrueHD, DTS-HD Master Audio, Atmos and plain FLAC. It plays surround through Apple's Spatial Audio renderer on AirPods, or across several USB DACs at once as one speaker system (Speaker Groups, in beta), measures your room with a USB microphone and corrects each speaker, shows every step between the file and your DAC, and tells you whether your "hi-res" files really are. It also plays to network players and AirPlay receivers, plays the music on your own server, and takes orders from a phone. Each of those is off until you turn it on.
Download the DMG or brew install --cask szeremeta1/tap/vespertine · macOS 14.4+ · Apple silicon and Intel
Signed and notarized · no account, no telemetry · Website · Press kit · Privacy · Sponsor
Vespertine stands with Ukraine. Please donate through United24, Come Back Alive or Razom for Ukraine, or see more ways to help.
- Speaker Groups (Beta): several DACs, one speaker system. Play one stream across two or more DACs as a single quad or surround setup, with a channel map, a room in three dimensions to place your speakers in, layouts from stereo to 9.1.6, and per-speaker level, delay and polarity. 5.1 and Dolby Atmos beds are placed on the speakers you actually have (tested with virtual devices only), and the signal path says exactly what that did to the samples. Off by default, and a beta: read what is and isn't verified.
- Room correction, measured in your room. Measure with a UMIK-1 or another calibrated USB microphone at one to five spots around your seat, and Vespertine fits filters to each speaker and lines up their delay and level. Stereo and 5.1 (5.1 is tested with simulated rooms and imported profiles, not yet in a real 5.1 room). It also imports REW and Equalizer APO filters and impulse responses. Off by default and bit-perfect while off. More below.
- Stereo and surround in one library. SACD 5.1, multichannel FLAC, DTS CDs, TrueHD and DTS-HD MA sit next to your stereo albums. A song you have in stereo and in surround or Dolby Atmos, on one album or across editions of it, is listed once, and your output picks the version that plays: surround on AirPods or a multichannel device, stereo on a stereo DAC.
- Disc images, played directly. SACD
.isoimages (stereo and multichannel, DST or plain) and DVD-Video LPCM (.isoor aVIDEO_TSfolder) play straight from the image, with nothing mounted or extracted. - Spatial Audio for your own files. Local surround plays through Apple's Spatial Audio renderer on AirPods (head-tracked where macOS provides it), with Atmos objects rendered from Dolby Digital Plus. Surround can also be exported as binaural stereo or multichannel ALAC.
- A signal path you can check. Decoding, resampling, mixing, gain and output format are shown for every track, read back from Core Audio. BIT-PERFECT appears only when the conditions in code are met, and when something changed the samples it says what. A built-in loopback test checks the software path sample by sample.
- Native DSD. DSF and DSDIFF from DSD64 to DSD512, including DST-compressed files, go out as DoP to DACs that take it (up to DSD256 on the DACs tried), or convert to high-rate PCM.
- Fake hi-res detection. Flags 16-bit audio padded to 24, likely upsampled masters and likely lossy sources, with the measurements behind each verdict.
vespertine-analyzeruns the same analysis on a Linux server next to your files. - Speaks your language. English and 44 more, from Ukrainian and Japanese to Amharic and Tamil, with right-to-left layouts for Arabic, Hebrew, Persian and Urdu and separate Simplified, Traditional and Hong Kong Chinese. A check on every change confirms each string exists in every language with matching placeholders and plural forms, and every screen was reviewed in every language for clipped or overlapping text. Corrections from native speakers are welcome.
Everything here is optional and off until you switch it on.
- Network Players (UPnP, beta). UPnP AV and OpenHome players on your network (moOde, Volumio, streamers) are listed as outputs. A player gets the original file. If the equalizer, room correction, ReplayGain or Vespertine's volume is in use, it gets 24-bit FLAC made after that processing, and the signal path says so.
- AirPlay Players (beta). AirPlay receivers are listed as outputs too, AirPlay 2 ones included. A receiver that asks for a password or shows a code is asked once, and the answer is kept in your keychain. A CD-quality file goes out as its own samples in ALAC; everything else is converted, and the signal path says which. An Apple TV 4K is sent 24-bit / 48 kHz for anything above CD quality. Receivers that show covers get the album art.
- Music Servers (OpenSubsonic, beta). The music on your own Navidrome, Gonic or Airsonic server gets its own section in the sidebar and plays through the same engine and signal path as a file. If the server converts a song, the signal path says "Converted by the server" and the badge is not BIT-PERFECT.
- Receive from Other Apps (beta). Control apps such as mconnect and BubbleUPnP list this Mac as a player and send it music, which plays on the output you chose with the same signal path as a file. Your own queue waits and comes back afterwards.
- A remote for your phone. Play, queue, browse and search from a phone, a tablet or any browser on your network, and add the page to a home screen to open it like an app. Phones pair with a QR code or a 6-digit code, each with its own key. Off by default, in Settings › Remote.
- Mac Sound Output (beta). A speaker group becomes an output of the Mac under the name you gave it, so a browser, the TV app or a video player can play stereo, 5.1 or 7.1 to it, with the volume keys working and Vespertine closed. Details and limits.
- Visualizer. A window of its own with a spinning record, spectrum bars, a spectrogram, club lights, a goniometer, an oscilloscope, and VU and PPM meters modelled on real broadcast, studio and tape-deck instruments. It reads a copy of what the DAC gets and never touches the sound. How it works.
- Transitions. Crossfade, Smart Mix or Live Set blend one song into the next. Albums, CUE sheets and SACDs always play gapless, and with the switch off every song plays as before, bit for bit. Details.
- Smaller things. A left/right balance for each output that changes nothing while centered. Spatialize stereo, which upmixes stereo to 5.1 for Spatial Audio or a multichannel output. A 5-second test that finds out whether your DAC takes native DSD. A Local Library switch for listeners who play only from a server.
Receive from Other Apps and Music Servers have been tested with Vespertine's own test control point and test server, not yet with real control apps or servers. Reports are wanted. Streaming in detail.
On iPhone and iPad. Vespertine for iPhone and iPad is a separate app on its own schedule. It is a remote for the Mac, and it can also play the Mac's library on the phone itself. It can set the Mac's equalizer, room correction, Spatial Audio and balance. Everything else on this page is about the Mac app.
This is a beta. It is off by default, behind one switch in Settings › Playback › Speaker Groups (beta), and with the switch off nothing about playback changes. The routing, the editor and the room-correction pieces are covered by automated tests and by opt-in tests against virtual devices. It has also been tested by hand on a real pair of DACs (a Fosi Audio ZD3 and a FiiO K11) in several configurations and settings. One earlier session on that pair, with a UMIK-1 and a probe tool rather than the app, measured how the DACs start. Reports from other DACs are very welcome.
Two or more USB DACs, played as one multichannel system: a Fosi ZD3 and a FiiO K11 as the front and back pairs of a quad, for example. Vespertine builds a private Core Audio aggregate of your DACs that exists only inside Vespertine (other apps can't see it or play into it) and removes it when playback stops. One DAC, the clock leader, sets the time; every other DAC is resampled by a few millionths to follow it, so the speakers stay together without the image drifting.
The room and the channel map below are cut from the editor of a demonstration group, which the app can make for screenshots. Its two outputs carry the names of the DACs this was tried on by hand, but no DAC was connected. The signal path further down was taken on two virtual audio devices (Vespertine Loopback and BlackHole 2ch) standing in for two stereo DACs. All three show the screens, not a measurement of real DACs.
Watch the demo (37 seconds, no sound): turning the switch on, creating a group from Vespertine Loopback and Vespertine Loopback B, editing the room drawing, then playing an album to the group. Both members are virtual audio devices standing in for stereo DACs. The recording used the 1.1.2 release with a development bundle identifier (ending in .dev) and an ad-hoc signature, which is what lets the group editor offer virtual devices. It shows the interface, not hardware timing or sound quality. The room has since become three-dimensional, as in the screenshot above.
- Channel map. Every channel of every DAC is assigned to a speaker (front left, centre, back right…). An Identify button plays three quiet bursts of pink noise (-30 dBFS) on that one channel of that one DAC, so you can check the wiring, and each channel has an Inverted box. Up to 16 output channels in total.
- Layouts from stereo to 9.1.6. Stereo, 2.1, 3.1, Quad, two pairs facing, ITU 5.1, 7.1, 5.1.2, 5.1.4, 7.1.2, 7.1.4 and 9.1.6, with the ceiling speakers at the ceiling. A new group starts on the layout that fits its channel count.
- Outputs you can change. Add, remove and reorder a group's DACs and pick its clock leader without starting over.
- Your room, in three dimensions. Drag the speakers and your seat to where they are in a room you can turn, zoom, or look at from above, the back or the side. Speakers have a height, the room has a ceiling and the seat an ear height. Each speaker's distance, angle, delay and level are worked out from it, so every speaker reaches the seat together. "Use the Drawing" turns that into a room-correction profile (delays and levels, no microphone needed), or measure the group with a microphone and the wizard also times each follower DAC against the leader.
- Per-speaker trims. Level (-40 to +12 dB), extra delay (0 to 40 ms) and a polarity flip for each speaker. A headroom guard lowers the shared preamp by the largest boost any channel can reach, so a trim doesn't clip a loud song.
- Surround placed by where your speakers stand. A song's channels go to the speakers by position: a channel with a speaker of its own plays there untouched; one between two speakers is panned between them with equal power (halfway is -3 dB each, as in ITU-R BS.775); one outside every pair goes to the nearest speaker at -3 dB. A 7.1 song on a group with one surround pair keeps both of its surround pairs there: the pair with no speakers of its own (the sides, on a group with a back pair) joins the other on the surround speaker of its side at -3 dB, and none of it goes to the fronts. Stereo, 5.1 and 7.1 files, and the 5.1/7.1 channel bed of a Dolby Digital Plus Atmos file, are all routed this way. The LFE is dropped by default where there is no subwoofer; you can instead send it to every speaker at a level, or to one speaker. Stereo on a quad plays the front pair only by default, or mirrors to the back at a level, or gives each pair its own stereo, or uses every speaker.
- Fill the room. A switch for a room where people listen from anywhere: every speaker plays the whole song or film. "Stage in front" keeps the sound in front at the seat (the fill is 3 dB down and 15 to 18 ms late); "Even everywhere" plays the fill at full level for the most even sound around the room. Nothing the group plays is bit-perfect while it is on, and it has not been measured or listened to on real DACs yet.
- A signal path that stays honest. The group is labelled BIT-PERFECT only when every channel the song uses lands on a DAC that is not resampled (the leader, or DACs sharing a hardware clock) and nothing changes the samples: no mixing, trim, filter or digital volume. Otherwise the signal path says what changed. A drift-corrected follower is always slightly altered; on virtual devices it measured 103 to 131 dB signal to noise, far below the noise of a real DAC and amplifier, but it is not bit-perfect.
- If a DAC goes away, the group either fades and pauses and names the DAC, or carries on with the DACs that are left (after a short mute) and folds the music onto them. It never carries on silently on a new clock in the middle of a song.
The signal path with a group playing. A 5.1 test file folded into the quad by speaker position: the leader's pair is bit-perfect and says it is the clock leader; the other pair is marked resampled, and the group as a whole is not labelled BIT-PERFECT.
Outputs and the channel map. The group's DACs in the order of their channels, with the clock leader, then every channel of every output against the speaker on it, with an Inverted box and an Identify button on each.
Limits, plainly:
- Beta. Tested by hand on one real pair of DACs (Fosi Audio ZD3 and FiiO K11). Hand testing is not a measurement: not yet measured from the app on hardware are a bit-exact leader, follower quality, a quad channel for channel, the Identify tone, a real unplug and one hour of playback. The script for the next hardware session is in the repo.
- Only the clock leader's channels can be bit-perfect. Start offsets between DACs were repeatable to about 21 µs in typical runs, but 3 of 49 drift-corrected runs came in about 3.5 ms late, unexplained, and a follower crept steadily behind the leader (+13.5 ppm, 0.81 ms a minute, in one pair). The wizard can measure that rate and trim for it, but the trim has not been confirmed on hardware over its full 10 minutes.
- PCM only: DSD is converted to PCM, Dolby and DTS bitstreams are decoded rather than passed through, and integer mode, "Match 16-bit sources" and head-tracked Spatial Audio don't apply to a group. On a group, Atmos plays its channel bed, not macOS's object rendering.
- It is for speakers around one listening seat, not whole-home audio: Bluetooth and AirPlay outputs can't join (they add 100 ms or more of changing delay), and a macOS built-in output can join but is never bit-perfect.
- In Vespertine the group has no volume of its own, and the volume keys don't control it: set each DAC's volume once, by hand. Published with Mac Sound Output (beta), for other apps' sound, it has a volume the keys set.
- The stereo-to-quad rows come from the routing rules, not a standard, and still need listening approval.
Speaker Groups in detail · Mac Sound Output (beta)
Measure your room with a UMIK-1 (or another calibrated USB microphone) from one to five listening spots. Vespertine plays sweeps through one speaker at a time and designs filters for each speaker from the measurement. They are cut-dominant: they take down the room's peaks and boost broad dips only gently. The same measurement sets each speaker's delay and level, so they all reach your seat together and at the same loudness.
- Targets: Warm (the default), Gentle tilt or Flat, for the bass only or the full range. Change the target and redesign the filters without measuring again.
- Import and export: REW and Equalizer APO filters (each speaker's delay, level and filters) in, Equalizer APO text out. Impulse responses (WAV, AIFF, CAF or FLAC) play by convolution.
- Stereo and 5.1 (5.1 is tested with simulated rooms and imported profiles, not yet in a real 5.1 room).
- Profiles: keep several for each output and switch between them, in a Guided or an Advanced view.
- See the change: the Room Portrait shows the speakers, your seat and each microphone spot as it's measured, and the Curve Reveal turns the measured response into the predicted one. A before/after switch compares them. "After" is predicted, not measured.
- Off by default. Playback is bit-perfect while it's off. While it's on, the signal path says ROOM CORRECTION.
Measured in a real room: a stereo pair on a Fosi Audio ZD3, measured with a UMIK-1 (its serial number is blacked out), with the Warm target. The thin lines are the measurement, the bold lines the predicted result, and the window scrolls on to each speaker's delay, level and filters.
Imported from REW (sample): a 5.1 profile imported as Equalizer APO text, showing what the correction does to each of the six speakers.
- Output: Switches your DAC to each file's sample rate, with exclusive (hog) mode, integer mode and an option to send 16-bit files as 16-bit. Remembers your DAC and falls back to the best connected output when it's unplugged. Lossless 24-bit / 48 kHz on AirPods Max over USB-C.
- EQ: A 20-band parametric EQ that imports AutoEQ and Equalizer APO presets per output. It's off by default, and when it is on the signal path says EQUALIZER.
- Library: Albums, artists, songs and genres with combinable filters, smart playlists in units like 24-bit and 88.2 kHz, gapless CUE-sheet albums, and favorites. Imports M3U and Apple Music playlists, play counts and favorites.
- Metadata: A tag editor that backs up every file before writing, with MusicBrainz and cover art lookup.
- Network shares: SMB, NFS and WebDAV, with caching and Keep Offline. Library scans step aside for the song you're playing.
- Everyday: Mini player, menu-bar extra, media keys, an optional Discord status with cover and format, ListenBrainz scrobbling, ReplayGain, a short setup assistant and automatic updates.
Formats: FLAC, ALAC, WAV, AIFF, DSF, DSDIFF, SACD ISO, DVD-Video LPCM, Dolby Digital, Dolby Digital Plus with Atmos, Dolby TrueHD, MLP, DTS, DTS-HD MA, DTS CDs, APE, WavPack (including DSD), TTA, TAK, Opus, Vorbis, Musepack, MP3 and AAC, the audio of Matroska and WebM files, plus CUE sheets (beside the file, or inside a FLAC).
All features, formats and screenshots
Bit-perfect playback is available in several Mac players. Vespertine focuses on a surround collection, a signal path you can inspect, and GPL source you can audit. Here's what each player's own documentation or source code says. "Not documented" means the vendor doesn't say either way; it is not a claim that the player lacks the feature.
| Vespertine | Apple Music (Mac) | Audirvana Studio / Origin | Roon | VeraVox | Colibri | Cog | foobar2000 for Mac | |
|---|---|---|---|---|---|---|---|---|
| Source code | GPL-3.0 | Closed | Closed | Closed | Closed | Closed | GPL-2.0 | Closed |
| FLAC / DSD files | Yes / Yes | No / No | Yes / Yes | Yes / Yes | Yes / Yes | Yes / Yes | Yes / Yes | Yes / No⁴ |
| Automatic DAC rate switching | Yes | No | Yes | Yes | Yes | Yes | Yes | Not documented |
| Exclusive (hog) mode | Yes | No | Yes | Yes | Yes | Yes | Yes | Yes |
| 16/24-bit PCM without sample changes | Conditions + tests | macOS 27: third-party tests¹ | Vendor claim | Vendor claim | In-app loopback test | Vendor claim | In-app indicator | Not documented |
| Local surround files | Yes² | Not documented | Yes | Yes | Yes | Not documented | Yes | Not documented |
| Apple Spatial Audio for local surround | Yes | Catalog Atmos only³ | Not documented | Not documented | Not documented | Not documented | Yes | Not documented |
| Hi-res file analysis | Yes | No | AudioScan | Dynamic range, loudness | Yes | Not documented | No (live spectrum only) | No (live spectrum only) |
| SACD ISO | Yes | No | Yes | No (DSF/DFF only) | Yes | No | No | No⁴ |
| Room correction | Measured in the app (USB mic), or REW / APO and impulse-response import | No | Via convolution (Origin: paid option) or Audio Units | Via convolution | No (by design) | Not documented | No | Not documented |
| Convolution / audio effect plugins | Yes (room correction) / No | No / No | Yes (Origin: paid option) / Audio Units | Yes / No | No / No | Not documented / Not documented | No / No | Not documented / Audio Units |
| Streaming services | No client for a commercial service; plays your own OpenSubsonic server and what UPnP control apps send (both beta, off by default)⁵ | Apple Music | Studio: yes; Origin: no | Yes | No (local + UPnP) | No (internet radio) | No (internet radio) | No (internet radio) |
¹ Apple Music on macOS 27 was bit-exact at 16/24-bit in third-party loopback tests reported on ASR. This is not a Vespertine measurement or a claim about every Mac configuration. Volume was at 100%; there is still no automatic rate switching, hog mode, FLAC or DSD on Mac. The thread was read again on November 17, 2026: it is a forum member's report made with an RME ADI-2 DAC's bit-test files, and nobody on this project has repeated it.
² Multichannel DAC output and receiver bitstream are experimental, untested on real receivers and multichannel USB DACs (one Sony TV over HDMI has been tested: see docs/DEVICES.md); reports wanted. Automated routing and carrier tests do not establish hardware compatibility.
³ Apple documents Spatial Audio with head tracking for Dolby Atmos tracks in the Apple Music catalog. It does not document spatial rendering of your own multichannel files.
⁴ Tested on foobar2000 2.26 (macOS 27.0.1), with no components: DSF, DFF and SACD ISO files fail with "Unsupported file format". DSF and DFF play only as 352.8 kHz PCM after installing FFmpeg separately and adding them to its FFmpeg Decoder Wrapper; SACD ISO still fails.
⁵ Music servers: your own server that speaks the open OpenSubsonic API (Navidrome, Gonic, Airsonic and others), tested against Vespertine's own test server, not yet against real ones. Receive from Other Apps: Vespertine has no code for any commercial streaming service. A UPnP control app that signs in to a service itself and hands over an ordinary http address can play through it; protected streams (Apple Music, Spotify, Amazon Music) can't. Tested against Vespertine's own test control point, not yet against real control apps. See docs/STREAMING.md.
Comparison sources and scope. Corrections welcome: open an issue.
PCM playback has been tested up to 768 kHz and DSD256 on a Fosi Audio ZD3, and up to 384 kHz and DSD128 on a FiiO K11 (developer reports). Playback has also been tested on AirPods Max over USB-C and Bluetooth, MacBook Pro speakers, and a Sony TV over HDMI in 5.1 (details). Device reports help extend that.
- Channel-for-channel output has been reported on one real device, a Sony TV over HDMI (5.1 and a 7.1 to 5.1 downmix, developer report without a capture; see Tested devices). Multichannel USB DACs and Dolby/DTS bitstream to receivers are untested on real hardware; reports wanted.
- Intel Macs are untested (the Intel build has run under Rosetta). DoP is reported to work up to DSD256 on the Fosi Audio ZD3 and up to DSD128 on the FiiO K11 (developer reports, no capture); DSD512 over DoP is untested on hardware.
- No audio effect plugins and no client for any commercial streaming service. Convolution is only part of room correction. Network Players (UPnP), AirPlay Players, Receive from Other Apps and Music Servers (OpenSubsonic) are beta and off by default. A network player or AirPlay receiver is never labelled BIT-PERFECT: the badge says what was sent.
- No DVD-Audio or DRM-protected Apple Music downloads. Dolby AC-4 and MPEG-H 3D Audio (Sony 360 Reality Audio) files are listed and named but not decoded, MQA plays as the plain FLAC it contains, and Auro-3D as ordinary surround PCM (no decoder for any of them; see docs/FEATURES.md). TrueHD Atmos and DTS:X objects are not rendered, and TrueHD and DTS-HD MA receiver passthrough is unsupported on macOS.
You need Xcode and XcodeGen (brew install xcodegen):
scripts/generate-project.sh && open Vespertine.xcodeprojRun the tests with swift test in Packages/VespertineKit and Packages/VespertineAnalysis (the analysis core also builds on Linux), or scripts/audit.sh for everything. docs/DEVELOPMENT.md covers command-line builds, a synthesized demo library, and the tools for checking playback on your own hardware.
Bug reports, device reports and pull requests are welcome. See CONTRIBUTING.md and the code of conduct.
GPL-3.0-or-later. See LICENSE. Third-party components and their licenses are listed in THIRD-PARTY-NOTICES.md.
Dolby, Dolby Atmos, Dolby Digital, Dolby Digital Plus and Dolby TrueHD are trademarks of Dolby Laboratories. DTS, DTS-HD Master Audio and DTS:X are trademarks of DTS, Inc. Apple, AirPods, AirPods Max and Spatial Audio are trademarks of Apple Inc. Vespertine names these formats and products only to describe what it plays and where. It is not certified, licensed or endorsed by Dolby, DTS or Apple, and it doesn't use their logos.
Vespertine was called Nocturne until version 0.6.0.







