# HeadMania.org Upsampler — 1 Million Taps Audio Processor > 1 Million Taps Audio Processor (iPhone/iPad app by DAN ALEXANDRU GHEORGHE.) - Source: https://appshunter.io/ios/app/headmaniaorg-upsampler/id6786069896 (this page in markdown: same URL + `.md`) - Developer: [DAN ALEXANDRU GHEORGHE](https://appshunter.io/developer/6786069898) - Category: Music, Reference - Price: Free with in-app purchases - Age rating: 4+ - Requires: iOS 15.0 · 265 MB - Languages: American English - Released: 2026-08-26 - Data updated: 2026-10-01 - User reviews in markdown: https://appshunter.io/ios/app/headmaniaorg-upsampler/id6786069896/reviews.md ## What is HeadMania.org Upsampler? 1,000,000-taps audio processor in your pocket. HeadMania UpSampler rebuilds your music with the same 64-bit filter engine as the desktop app, then feeds a USB DAC at the highest rate it accepts. On an M-series iPad, four million taps. New: Neuromancer, an experimental neural renderer that runs on the Neural Engine and shapes the attack of each note as the music plays. Free. What you need HeadMania UpSampler does not play through the iPhone or iPad speaker. It needs a real output: • A USB-C or Lightning DAC — a plain USB-C-to-3.5 mm dongle counts. • Or a UPnP/DLNA network DAC or streamer on your Wi-Fi. Bluetooth and AirPods are not supported. Room correction needs a microphone. To receive from a control point, both must be on the same Wi-Fi, and you must allow Local Network access when iOS asks. THREE WAYS TO FEED IT - Your own files on the device: FLAC, WAV, AIFF, ALAC, CAF, MP3 - Lossless internet radio — 14 stations whose operators have given written permission to carry them - UPnP/DLNA receiver — send to it from mconnect, JPLAY, Audirvana or any control point OUTPUT PCM to 705.6 / 768 kHz over USB, if the DAC supports it. THE FILTERS Eight families, each at several kernel lengths, so you can hear the difference rather than take it on trust: Organik (minimum phase, no pre-ringing), Organik Touch and Organik Pulse (drum hits take a short path, the sound around them keeps the long Organik reconstruction), Linear (linear phase), Mixed Phase (between the two), Organik-Adaptive, Transient-Perfect (optimised for step response) and Kinetic (polyphase). Lengths run from 6,000 taps to 1,000,000 on iPhone and 4,000,000 on an M-series iPad. The tap count shown is the measured length of the kernel actually built. If memory runs short or playback starts to starve — sustained heat can do it — a lighter filter of the same family takes over mid-track, and the app says so. NEUROMANCER A network with 8.5 million learned values reshapes the attack of notes as the music plays: the kick, the bass note, the snare hit come forward by a fraction of a decibel to about two, and little else changes. It is not an EQ. Four choices: Neuromancer, Awakened, Revelio and Revelio's Pulse. Compare with Original. It is demanding: it needs iOS 27 and was tested on iPhone 18 Pro Max and iPad Pro with M5. The app times each model on your device and keeps Neuromancer off where it cannot keep up. Expect more battery use and warmth. Playback starts about three seconds later. Trained on openly licensed multitrack recordings, credited in About. ROOM CORRECTION Your room does more to the bass than any filter does. Record a test sweep with a microphone and the app finds the resonances the room adds, then places parametric cuts at them, each sized to its measured height and adjustable by hand. Where a note keeps ringing it can also shorten the decay, previewed first and marked experimental. Bass only, one seat; no equaliser fixes reflections. Without a microphone it can design from a floor plan. Room correction is still in testing and says so. WHAT IS FREE, PERMANENTLY Every filter below 500,000 taps, at every output rate. Neuromancer. All three input paths. Measuring your room, with the graph, the editor and saving. No account, no time limit. SIGNATURE EDITION One optional purchase adds the presets from 500,000 taps upward (on an M-series iPad up to 4M) and hearing room correction applied. One payment, no subscription. A 30-day trial comes first; when it ends the filter moves to the deepest free preset of its family and playback continues. PRIVACY No account, no analytics, no crash reporting, no telemetry. Neuromancer runs on the device. The app listens on the local network so control points can find it, and connects out only to radio stations you choose, the AutoEq database when you search it, and the App Store. REQUIREMENTS A USB or network DAC and iOS 15 or later; Neuromancer needs iOS 27. A renderer and a player, not a music service. ## Pricing and in-app purchases Base price: Free. **One-time purchases** | Purchase | Price | Type | | --- | --- | --- | | Signature Edition | $29.99 | one-time | | 30 Days Free | Free | one-time | ## Version history (last 4 releases) ### 3.1.5 — 2026-09-29 Neuromancer comes to iPhone and iPad. It is a neural renderer: a network with 8.5 million learned values that shapes the attack of each note as the music plays, before the filters do their work, and changes little else. It runs on the Neural Engine, entirely on the device, and it is free. • Four choices: Neuromancer (bass-note attacks), Awakened (aimed at attacks only, sustained notes left alone), Revelio (drum attacks), and Revelio's Pulse (snare, stick and hi-hat attacks; the strongest, up to about 2 dB). • The changes are small on purpose. Switch to Original to compare, and choose by ear. • It is experimental and demanding. It needs iOS 27, and it was tested on iPhone 18 Pro Max and iPad Pro with M5. It keeps the Neural Engine and processor busy while it plays, so expect more battery use and a warmer device. The first time you open this version, the app times each model on your device by itself; where a model cannot keep up, Neuromancer stays off. • Playback starts about three seconds later, by design. Radio: Radio Club 80 — Euro joins the lossless stations, with its operator's written permission. Also new: a short intro when the app opens (Skip passes it), and an iPad now appears as "HeadMania.org iPadOS" in control points. ### 3.0.1 — 2026-09-12 New filters: Organik Touch and Organik Pulse. Attacks such as a drum strike or piano hammer take a short filter path while the sustained sound around them keeps the long Organik reconstruction; Pulse separates more selectively around detected hits. Both are preview filters with extra processing and playback delay — choose by listening. Room correction, further along: • Measurement is a guided six-capture flow, with a floor plan optional. • Lingering bass: resonances that keep ringing after a note are detected and can be shortened where the analysis finds a correction that passes its checks. Ringing reduction is experimental and the prediction is not a measured result. • Fullness keeps more of the bass balance you prefer; recalculating from saved recordings needs no new measurement; the previous design can be restored. • The correction table lists response and ringing filters side by side, and the room graph shows detected and remaining ringing. • Correction can be switched off before playback starts, and a floor-plan draft survives switching tabs. Radio: Pure Lounge Radio and Pure Classix Radio join the lossless stations, with their operator's written permission. Fixes: the app reconnects to its engine after a long stay in the background instead of showing "Reconnecting…"; the first screen no longer stays blank on a slow launch; exported room files can be picked when importing. Room correction is still in testing and not fully validated. ### 3.0.0 — 2026-09-03 New: room correction. Sweep your room with a microphone, see its bass resonances on a graph you can edit, and cut them. Measuring is free. Room correction in HeadMania works only in the bass, where the room dominates what you hear, and it is applied as a set of parametric cut filters (minimum-phase biquad notches). It is not a FIR convolution and not a full-spectrum "smoothing". How the filters are arrived at: 1. Measurement. A logarithmic sweep is played through the speakers and recorded with the microphone (Farina deconvolution). With several positions around the listening seat the responses are averaged; with a single position, only the bass is corrected. 2. Mode detection. From the averaged curve we extract the room's own one-octave trend and look for the peaks that rise above it. Those are the room's resonances, each with its frequency, height and width (Q). 3. The filter. For each mode, a notch whose zero is placed on the resonance's pole, with the Q matched to the measured width ("matched modal cancellation", Mäkivirta et al., JAES 2003). The aim is not just to lower the level of the peak but to cancel the resonance itself, meaning its ringing in time, which is what muddy bass is. Below the room's Schroeder frequency, magnitude and decay are the same phenomenon, so a correctly matched notch is also the time-domain correction. 4. The limits. Cuts only, no boost. A filter may not dig a hole below the room's trend, and the sum of the cuts is bounded by a headroom budget so no clipping can occur. The resulting filters are shown on the graph and can be adjusted by hand. What it does not do: it does not correct reflections or speaker-boundary interference (those cannot be fixed with EQ from a single position), it does not work above the bass region, and it does not yet apply a separate decay filter. Alternatively, without a microphone, the same filters can be designed from a simulation of the room (FDTD) based on the dimensions and speaker positions you describe. ### 2.9.43 — 2026-08-27 No release notes. ## Related topics [upsampler](https://appshunter.io/ios/topics/upsampler) --- *Data collected daily from the US App Store and indexed by [AppsHunter](https://appshunter.io/). User reviews are verbatim App Store reviews. Ratings, prices and chart positions refresh continuously; this snapshot is from 2026-10-01.*