00Why offline

Every DAC upsamples. It has to — reconstructing an analog waveform from 44,100 samples per second requires an interpolation filter. But a DAC chip does it in real time with a filter of a few hundred to a few thousand taps. AuraEngine removes the real-time constraint entirely: it renders the track offline through FIR filters of 5 thousand to a billion taps, designed in 128-bit arithmetic, applied in end-to-end 64-bit float. The DAC then receives a 352.8/384/705.6/768 kHz (or, since 1.8.0, 1411.2/1536 kHz) waveform that is already reconstructed — its own filter has almost nothing left to do. How long the filter is, is a setting, not a quality tier: on music a 5k and a 30M render agree below 20 kHz, and what length changes is how steep the wall at the cutoff is (the stop band is deepest at 30M; longer filters make the wall steeper, not the floor deeper) (measured).

Since 1.5.0 the same chain also plays. The player prepares the source stages for the whole track ahead of time, runs the convolution on the CPU — or on the graphics card when the CPU cannot keep up — and shows which stages are already in the sound. It plays internet radio the same way, as the stream arrives. The converter stays the place where nothing is rushed: every stage at full length, and the written file verified against the render (the player, radio).

01The signature: Hybrid-Phase

Invented for this project. Every FIR filter forces a trade: linear phase keeps the stereo image coherent but pre-rings before every attack; minimum phase hits clean but warps inter-channel timing. Fixed "intermediate-phase" compromise filters carry a little of both flaws, everywhere, all the time. AuraEngine refuses the trade — it renders the track twice in full and switches between the two renders per-transient.

What the detector keys on, precisely. It measures the rise of percussive energy between analysis frames — the start of a sound. It does not measure whether the filter would ring on this material; it knows nothing about the transition band. On music the two coincide, because an attack is both a beginning and a broadband event. On synthetic material they come apart: a steady square wave with arbitrarily steep edges produces no trigger at all, because nothing begins. So the honest name for the stage is minimum phase on attacks, and there is a known blind spot in the first ~24 ms of a file — both are written up, with the numbers, in doc 06 §1.

02The signal path

One file, start to finish. Stages marked optional are user toggles; everything else always runs. The entire CPU sample path is 64-bit float — there is no f32 truncation anywhere between decode and dither.

1

Decode converter/decode.rs · Symphonia

FLAC, WAV, MP3, OGG and AAC/M4A are decoded by the pure-Rust Symphonia library straight into f64 — HE-AAC by the FDK AAC decoder built into the program, at its full rate, and a chained Ogg file to the end of its last link. Integer PCM keeps every bit — a 24-bit master divides by 2²³ exactly — and a lossy decoder's output is kept as it comes, including the samples that sit above full scale, which MP3 and AAC legitimately produce. Nothing is clipped on the way in; the output ceiling decides what happens to those peaks. MP3/AAC encoder padding is trimmed using gapless metadata.

Why it matters: the chain starts bit-exact, and it starts with the whole signal. A 24-bit FLAC enters with all 24 bits intact; an MP3 that peaks at +7 dBFS enters at +7 dBFS.

f64 decode, no clampgapless trimno external tools

2

DC block pipeline/prepare.rs

Per-channel mean removal (default) or an optional 2 Hz single-pole IIR high-pass with the exact closed-form pole r = e^(−2π·2/fs).

Why it matters: a fraction of a millivolt of DC offset would make a 30M-tap filter ring at infrasonic frequencies and poison the true-peak measurement.

static meanor 2 Hz IIR HPF

3

Declip dsp/lab/declip.rs optional · on by default

A master pushed into the rail came back with its peaks flattened. The flat top carries no information about how high the wave was going; the flanks either side of it do. Declip finds those runs and rebuilds the arc across them by constrained autoregressive interpolation, with the rail treated as a lower bound rather than a value.

The flanks fix the arc; the rail fixes its floor.

The longest flat run in a real track — 57 samples, 1.29 ms — and the arc put back over it, peaking 6.02 dB above the rail.

Why it matters: it refuses far more often than it fires. A run shorter than 17 samples is a limiter working, not a clipper, and the stage stands down and says so — on a 2014 Santana master, twelve tracks, not one run reached the threshold. The output ceiling takes the rebuilt peaks back down, so a declipped file arrives quieter: level-match before comparing. Since 1.5.0 it leaves MP3, AAC, Vorbis and Opus sources as decoded: their samples above full scale are the codec’s, not a clipped master’s.

DC tagrail as a lower boundgated at 17 samples

4

Intersample peak correction dsp/lab/isp.rs optional · on by default

A waveform is not only its samples. Between them the reconstructed signal can go higher than any sample does, and higher than full scale. ISP scans at 4× with a Lanczos kernel and applies the smallest correction that clears the overs. At the end of the chain its second half, a limiter with ~1.5 ms dips, trims the overs the reconstruction itself makes, so the whole file does not come down for a few milliseconds of peaks.

Twenty-three samples of a real track: every sample sits at or below full scale, and the signal between them reaches +3.68 dBTP.

Why it matters: this is where a DAC or a downstream encoder clips a file that never exceeds 0 dBFS. Spans Declip has just rebuilt are left alone — they are deliberately above the old rail, and holding them down is the output ceiling’s job, not this one’s.

ISP tag4× Lanczos scanskips declipped spanslocal output limiter

5

Subsonic filter pipeline/prepare.rs · process.rs optional · on by default, 15 Hz

Runs here, after the repairs above, so a high-pass cannot tilt the flat tops they read. A linear-phase FIR high-pass at the source rate: a Kaiser-windowed sinc low-pass, spectrally inverted, so the response at DC is exactly zero. Flat within 0.01 dB from the corner up — 20, 15 or 10 Hz — and at least 100 dB down from half the corner to DC; 29 813 taps at 44.1 kHz and 20 Hz. Since 1.3.3 the same filter runs a second time at the output rate, after the intersample-peak limiter at the end of the chain: its short gain dips modulate the bass and would put a shelf of infrasound back under the corner.

Why it matters: subtracting the mean removes a constant offset and nothing else. A slow wobble goes through it untouched — one CD in the Audio Science Review thread carries a component near 1 Hz at about 1 % of full scale. This is protection for speakers, not a sound improvement. Linear phase is the point: above the corner nothing moves in level or in time. It costs look-ahead — half the filter — which the converter and the player’s preparation have for free, and a live radio stream pays as a short wait before its first sound.

SUB20 taglinear phase≥100 dB below F/2flat from the cornersecond pass after the limiter

6

Headroom dsp/true_peak.rs

The true-peak ceiling the finished render is normalised to: Off leaves the shipped −0.5 dBTP, and −0.5 / −1.0 / −3.0 dB name their own. Nothing is scaled before the filter.

Why it matters: it is a ceiling, not a target — a file already below it is passed through untouched. Until 1.2.3 this was a gain applied before the convolution, which the absolute ceiling at the far end then cancelled exactly: the control did nothing on any loud master.

Adaptive Headroom (on by default) decides whether to honour that ceiling on this file: it keeps the shipped −0.5 dBTP when the source peak already sits below the target, or when ENOB says the master is a clean high-bit-depth one, and otherwise lowers it as asked. It is a veto on the setting rather than a replacement for it, so with Headroom on Off there is nothing to decide and the stage does not run.

a ceiling, not a gainAHR tagneeds Headroom set

7

Adaptive Apodizer v3 — source forensics converter/apodize.rs optional · any source rate

Masters carry the fingerprints of every filter that ever touched them. The v3 analyzer reads those fingerprints instead of guessing: a time-domain detector isolates the near-Nyquist band, finds the strongest attacks across the track and looks for high-frequency energy rising before them — the signature of a linear-phase brick wall — then measures the ring's exact frequency (per-burst FFT, median across attacks) and places a corrective minimum-phase Kaiser lowpass just below the measured edge, with depth chosen by severity (β=24 strong / β=14 mild). A Welch spectral-cliff detector unmasks fake hi-res — upsampled 44.1/48 kHz masters inside 88.2k+ containers are treated against their original Nyquist — and a mirror-image alias probe catches bad upstream resamplers whose spectral images defeat every other test. A cliff with no pre-ring means the source filter was minimum-phase or already apodized: diagnosed in the log, audio untouched. Static presets (Gentle 20 kHz / Moderate 19 kHz / Strong 18 kHz) are available too.

Why it matters: upsampling scales flaws along with music. Healing the source's pre-ring first gives the big FIR a clean canvas — and tracks that don't need it pass through untouched. Since 1.5.0 it also stands down on a lossy file whose codec has already cut everything below where its cutoff would go: a filter there would only turn the phase.

AA tagmeasured ring cutofffake-hi-res unmaskingmirror-alias probemin-phase · zero pre-ring

8

Upsampling — two paths converter/process.rs

The engine has two resampling architectures, selected by the Polyphase FIR Resampling toggle. Both end at the same output stage.

Standard path any ratio · Rubato + FIR post-filter

• Rubato SincFixedIn<f64> upsamples first: 512-tap sinc, 512× oversampling, cubic interpolation, ≈−180 dB stopband.
• The million-tap FIR then runs at the output rate as a spectral-shaping post-filter — partitioned overlap-save, block 32768, Kahan-compensated frequency-domain MAC, parallelized with Rayon.
• Filter blob missing? The post-filter is skipped with a logged warning — Rubato alone is still a first-class resampler.

Polyphase FIR path integer ratios · the filter IS the resampler

• The same million-tap filter is decomposed into L sub-filters (h_k[m] = h[m·L+k]) that run at the source rate, in parallel on all CPU cores.
• Anti-imaging cutoff sits exactly at source Nyquist: fc = src_rate / (2·out_rate). Each source rate takes the filter made for its own ratio — 96 kHz to 384 kHz is ×4 — so a hi-res file keeps what it holds above the CD band.
• Output length is exactly input × L — no resampler padding at all. Non-integer targets snap down (44.1 kHz × FS8 → 352.8 kHz).
• Filter blob missing? Hard error — this path refuses to degrade silently.
• Multistage, the default since 1.7.0 (5M taps and longer, ×2/×4/×8/×16, and ×32 since 1.8.0): only the sub-filter that lands on the source's own samples runs — once, at the source rate — and a short exact interpolator (≥ 210 dB) does the rest. Several times less CPU; the difference to single-stage is −204…−220 dBFS. doc 21
• The rungs above 30M (100M, 180M, 300M, 500M, 1G) exist only as branch 0 files — the one sub-filter this path runs — so they play only here. doc 12 §11

9

Hybrid-Phase engine hybrid_phase.rs · hpss_native.rs · pipeline/hybrid_mixer.rs optional · ~2× time

Linear-phase filters preserve the stereo field but pre-ring before transients; minimum-phase filters hit hard but smear inter-channel phase. AuraEngine runs both convolutions in full, then a native-Rust HPSS (harmonic/percussive separation) onset detector builds a transient envelope: STFT with time/frequency median masks, 8-band spectral flux, adaptive gate, envelope follower with lookahead. During sustained material you hear the linear-phase branch; one to two analysis frames before each detected attack (~12–23 ms at 44.1 kHz) the engine switches to the minimum-phase branch — at a zero crossing, with a 1 ms fade and a 20 ms anti-chatter hold. The trigger is the start of a sound, not a measurement of where the filter rings.

Why it matters: the switch is stereo-linked — one switching plan computed from the mid signal, applied to both channels at the same sample. Channels never de-phase against each other, so the image stays rock solid while attacks stay clean. The minimum-phase branch is aligned using band-weighted group delay (200–6000 Hz) so both branches sit sample-exact at the perceptual core of the music.

HP tagstereo-linked switchzero-crossing + 1 ms fadeHPSS onset envelope

10

TFS phase dsp/lab/tfs.rs optional · on by default

A different way to spend the phase budget: linear phase below 1.5 kHz, minimum phase above 4 kHz, blended across the band between. The low end keeps the exact timing a linear-phase filter gives it, while the treble is rendered with no pre-ringing ahead of a transient.

Why it matters: the magnitude response is not part of the trade — it matches the linear-phase render to ±0.000003 dB across 20 Hz–20 kHz. Only the phase differs, which is why an ordinary frequency sweep cannot see this stage at all. It is derived from the linear and minimum pair of your filter size on first use and cached — at 30M taps about 12 s and 4.5 GB of RAM, once; since 1.5.0 on a transform twice the filter’s length, which takes its stopband to −193 to −231 dB. It replaces Hybrid-Phase for the file, and the two cannot both run.

TFS tag±0.000003 dB magnitudederived once, then cached

11

Crosstalk cancellation dsp/lab/xtc.rs optional · off by default

Over speakers, each ear hears both of them. XTC cancels what each speaker sends to the wrong ear, so the stereo image is no longer bounded by the speakers themselves. The filter is built from the listening triangle you measure with a tape.

Computed from a symmetric free-field model of the triangle, not measured in a room. Read the right-hand panel first: 20 mm of head movement costs about half the cancellation.

This is the one stage that deliberately colours the signal. Everything else here is a repair or a rendering choice; this rewrites the file for one chair. It will not switch on until the geometry is entered, and the same check is repeated on the Rust side so a hand-edited setting cannot slip past — a canceller built from guessed distances does not cancel less, it cancels the wrong thing. Speakers only, never headphones.

XTC tagneeds a measured triangleone chairnot transparent

12

Album level converter/album.rs optional · on by default · 1.5.0

The true-peak ceiling below is one gain per file, and how deep it cuts depends on what the chain did to that file — so converted one by one, the tracks of an album can come out with their balance changed. The queued files that share a folder and an Album tag are converted as one album: before the first is written every track is prepared and the cut it will need is estimated, and every track takes the deepest of them, in front of its output limiter.

Why it matters: on a 12-track album the spread against the CD went from 6.15 dB to 0.02. The price is level and a wait: an album comes out as quiet as its most demanding track needs, and no track is finished before all of them are prepared. A file without the tag, or the only queued track of its album, converts exactly as before. No filename token — the level goes into the log and an AURA_ALBUM_GAIN tag.

ALBsame folder + Album tagone gain per albumconversion only

13

True-peak limiter converter/dsp/true_peak.rs

4× oversampled intersample-peak scan using an 8-tap Lanczos-4 polyphase sinc interpolator (reflective edge padding, every sample inspected), measured on what the output limiter and the subsonic guard have left. If the reconstructed waveform would exceed −0.5 dBTP (or the Headroom setting), one single linear gain brings the whole file under the ceiling — with Album Level, the album’s shared gain. Quieter material is left bit-exact — no gain is applied at all.

Why it matters: loudness-war masters routinely hide +1…+3 dB intersample peaks. Without this stage they would clip in the DAC's reconstruction filter — as harshness on every loud passage.

ITU-R BS.1770-style−0.5 dBTP ceilingbit-exact when quiet

14

Dither & noise shaping converter/dsp/dither.rs

24-bit TPDF dither with two independent random streams — one per channel, so the dither floor cannot correlate between L and R. At output rates ≤48 kHz a 9th-order Wannamaker psychoacoustic noise shaper pushes the quantization noise where hearing is least sensitive. Since 1.8.0 every rate from 88.2 to 1536 kHz has a shaper of its own too — a minimum-phase error-feedback filter of up to order 16 with the least noise in 0–20 kHz at a bounded total (at most +10 dB over plain TPDF): the noise in the audio band falls by 13–15 dB at ×2, 31–35 at ×4, 57–59 at ×8, 61–62 at ×16 and 63–64 at ×32. A 16-bit player device gets the dither and the shaper on its own grid. Samples are clamped to ±(1−LSB) after quantization.

Why it matters: proper dither converts quantization distortion into benign noise — reverb tails and room decay stay intact instead of dissolving into correlated grit.

TPDF 24-bitper-channel RNGWannamaker-9 ≤48 kHzper-rate shaper >48 kHz

15

FLAC encode converter/encode.rs · native

Encoded to 24-bit FLAC in-process — no external tool and no subprocess. Frames are written to disk as they fill, so a 90-minute file costs the same peak memory as a three-minute one. The predictor order is searched per block, and the settings were chosen by measuring real converted material rather than by assumption. Cancelled or failed encodes delete the partial file.

24-bit FLACpure Ruststreamingoutput next to source

✓

Bit-perfect verification converter/utils/verify.rs

The file just written is decoded back and compared sample-by-sample against the DSP buffer, with a ±2 LSB tolerance at 24-bit. Pass → ✓ VERIFIED badge. Fail → the file is renamed _UNVERIFIED.flac and flagged in the UI.

Why it matters: this closes the loop. You don't have to trust the pipeline — the pipeline checks itself, on every single file.

every file, every time±2 LSB

03Choosing a path

Both paths share every preparation and output stage. The difference is who does the sample-rate conversion.

04Optional GPU acceleration without precision loss

A graphics card is not required: the CPU path plays and converts everything on its own. When one is there, it takes the convolution without losing precision. The standard objection to GPU audio is that consumer GPUs compute in f32 — a ~−150 dB noise floor that a 30M-tap convolution would happily amplify. AuraEngine's answer is double-single (DS) arithmetic.

05FIR upsampling on modest hardware

30-million-tap upsampling on a CPU, without a graphics card. Since 1.7.0 the long filter runs once, at the source rate, and a short linear-phase interpolator — designed to 218 dB, stop band at least 210 dB — takes the result the rest of the way to ×2…×32: multistage (cascade) upsampling. The long filter still decides the wall, the phase and the ringing; the work per output sample falls several times (how it works and how it was checked).

Measured on an Intel Core i9-14900K. The four efficiency cores rows pin the process to four of its E-cores, as a stand-in for a modest laptop CPU — not a real laptop. Source: 44.1 kHz CD material; filter: 30M taps (counted at the output rate), linear phase.

Player: CPU per second of music

On four efficiency cores a 30M ×8 stream now runs about 35 times faster than real time (about 7 times single-stage). In a real window, 30M ×4 went from one full core to a quarter of one.

Player: from the press to the first sound

The time until the first 0.3 s of output is ready, with XTC and the subsonic filter on; opening the device not included; median of three; 1.8.1 code, 2026-10-24. Playback does not wait for a conversion: the track is decoded and its source stages run ahead of what you hear, and the stages that need the whole track switch in as they get ready.

Converter and accuracy

The last bit of a 24-bit file is at −144 dBFS; the difference sits about 66 dB below it. Not bit-identical — rounded to 24 bits, one sample in 20 000 to 90 000 differs by one LSB — but the wall (−6 dB at 21 050.01 Hz), the phase, the ringing and the delay are unchanged, and in a blind A/B the listener could not tell the two apart. Filter length stays a setting, not a quality tier; multistage makes the long filters cheap enough to choose on an ordinary computer. It applies from 5M taps up; 1M (the default on a fresh install), 5k, custom filters and ×3/×6 stay single-stage.

06Filter design: 128-bit offline, f64 at runtime

The converter deliberately cannot generate filters at runtime. Coefficients come from fir-optimizer, a Python design suite that builds each Kaiser-windowed (β=14) sinc prototype in 128-bit arithmetic (mpmath, 38 significant digits), normalizes DC gain with extended-precision summation, and ships the result as f64 .npy blobs.

07The laws of the DSP core

From DSP_MANIFESTO.md — the axioms every commit is audited against. Violating one degrades the device class; that is the review bar for this codebase.

08Verification & the audit trail

09One engine, two ways: convert or play

The question is not offline against real time. Every file in the list has ▶ beside C, its convert button, and both run one engine and one chain — the same filter blobs, phase engine, rack and output stages, from the same code — which does not quietly give up sound for speed. The converter has no deadline. The player prepares ahead what a track needs whole and switches each stage in as soon as it is ready; once every stage is in, what it sends to the device matches the converted file to −120 dB (measured on 30M taps, FS×8, TFS, XTC). When the machine cannot keep up, it moves to the graphics card, then to fewer taps, and says so. A live stream has no future to read ahead into: only Declip and the Adaptive Apodizer stand down there, and the interface shows it.

Why convert, when the player matches it? Any filter on any machine; one level for a whole album; a file that plays anywhere, with no PC; no CPU and nothing to keep up with while you listen; and a file checked against the render it came from. Why play? To hear a setting before an evening goes into converting; to listen at once, without disk space for hi-res copies; and radio, which cannot be converted.

Every control, key and menu: doc 19.

10Internet radio through the rack

F / R at the head of the list switches it to the radio: featured stations, your favorites and recent ones, a search of the Radio Browser catalog — tens of thousands of stations by name, genre, country, language and format, sorted by votes, plays or quality and filtered down to lossless — or any stream's address. MP3, AAC, HE-AAC, FLAC, Vorbis and Opus over HTTP, .pls/.m3u playlists and HLS, all decoded inside the program, at 44.1 or 48 kHz and their multiples.

Upsampling cannot add what the codec removed: a 128 kbps MP3 stream stays a 128 kbps MP3, played through a better reconstruction filter. Stations that stream FLAC are where the rack has the most to work with.

11The ASIO input

F / R / I — the switch at the head of the list has a third view. I is the input: Aura Player ASIO, a virtual ASIO device. Choose it as the output in foobar2000, JRiver, HQPlayer, Roon or any player that can play to an ASIO device, press play, and the music goes through Aura Player's rack to your DAC — at the output rate, or bit-perfect if you switch that on. Windows only.

Spotify, or any Windows program

A program with no ASIO output of its own can reach the input through two programs from VB-Audio — a Hi-Fi Cable and an ASIO Bridge. They are third-party programs (donationware, Windows); AuraEngine does not include them, and how they behave is theirs.

1. Install VB-Audio Hi-Fi Cable & ASIO Bridge: run the setup as administrator, then restart Windows.
2. Install the driver in Aura Player's I view, if you have not.
3. In Windows sound settings, or in the program (Spotify and others), play to Hi-Fi Cable Input.
4. In ASIO Bridge, choose Aura Player ASIO and press ASIO ON.
5. Play. The view shows VBCABLE_AsioBridge; press Listen if it does not switch by itself.

The chain this was heard on: Spotify → VB-Audio Hi-Fi Cable (48 kHz) → VB-Audio ASIO Bridge (32-bit) → Aura Player ASIO → the rack → the DAC. The driver is GPL-3.0-only, the protocol between it and Aura Player MIT; the player stays under PolyForm Noncommercial.

12The analyzer

Ctrl+Shift+A opens a window on what you hear now; a row's right-click menu opens one on the file, or on its converted copy. It measures three signals side by side: S, the source; B, the source after the rack's repair stages, before the filter; O, the output of the whole chain, from a pass over the entire track.

The whole-track passes run only while the analyzer is open, so playback without it costs no extra CPU. On a radio stream it measures the stream and the output live — for the song playing and for the whole session — and shows how far the stream’s spectrum really reaches: an MP3 at 128 kbps ends near 16 kHz, a lossless stream reaches half its sample rate. Its views keep what has played — about three minutes of the spectrogram, 30 minutes of the loudness and of the waveform — and follow the live edge until you zoom in or look back.

13Scenes that draw the music

In listening mode the player can draw what it plays. A scene is a GLSL shader in an .aura-vis file; 19 come with the app, and a studio edits them, tunes their settings with sliders and takes Shadertoy's multipass shaders, keeping each one's author, link and licence. The player renders a few seconds ahead of what you hear, so a scene can see a kick coming and have it land as it sounds.

14Get it

Everything above needs filter coefficients, and those are the big part of the download — the full set, with the rungs above 30M, runs to tens of gigabytes. So the app ships four ways on Windows, each with its filters already inside (Linux and macOS: the cards above; the filters download in the app). Unzip, run aura-engine.exe, drop a track on the window: ▶ plays it, C converts it. The app opens on the filter that came with it: there is no setup step, and no way to land on a setting the package cannot serve.