Sound, in focus.
Measurement-driven DSP correction for specific playback systems.
Built for RootlessJamesDSP.
Overview · Profiles · Architecture · Quick start · Documentation · Validation · Limitations
HyperAudio publishes two device profiles, but their deployment guidance is not identical. The KZ EDX Pro uses the identity VDC → minimum-phase IRS → EEL2 chain. The Redmi 12 directory retains the frozen V6 three-file assets, while its current Speaker setup guide documents a separate standalone EEL2 PEQ configuration. These Speaker configurations are alternatives and must not be combined.
This is not a general-purpose preset collection. Profiles, measurements, implementation details, and test claims are scoped to the hardware and evidence actually available.
| 2 device profiles | 3 files per chain | 48 kHz IRS host rate | 1 tonal-correction stage |
|---|---|---|---|
| 2 device profiles | Profile-specific assets | Verify the running rate | No duplicated EQ |
See methodology for the evidence model and processing rationale.
|
8192-tap · 48 kHz IRS Important configuration split: the guide is for standalone Liveprog PEQ + limiter. It explicitly replaces the older VDC → IRS → limiter chain; do not load the VDC or IRS with this guide. The frozen V6 three-file assets remain in the repository and are documented separately in the technical report. |
2048-tap · 48 kHz IRS · 24 Ω · 112 dB/mW Selectable Effects or Limiter-only paths. The IEM voicing is a labelled design choice, not a measurement-derived target. |
The following is the three-file architecture used by the IEM profile and the frozen Speaker V6 asset set. It is not the current standalone Speaker PEQ guide configuration.
flowchart LR
A[Source] --> B["VDC<br/>Identity filterbank"]
B --> C["IRS<br/>Minimum-phase FIR<br/>Tonal correction"]
C --> D["EEL2 Liveprog<br/>Effects / Limiter"]
D --> E[Output]
Important: Tonal correction lives in the IRS. The VDC assets are identity filterbanks. Do not load the same EQ in multiple stages; doing so applies correction more than once.
- Choose the correct profile guide first. The IEM guide uses the three-file chain below. The Speaker guide uses standalone Liveprog PEQ and explicitly excludes its VDC and IRS files.
- For the IEM chain, set the host rate to 48 kHz, verify the running rate, and set convolver preamp to 0 dB.
- Disable other EQ, bass enhancement, and OEM sound effects that would stack processing.
- Load the IEM chain in order: DDC (
.vdc) → Convolver (.irs) → Liveprog (.eel). For the Speaker, follow its standalone guide instead. - Start at low volume and confirm the selected profile and processing mode.
|
Redmi 12 Speaker |
KZ EDX Pro |
Open HyperAudio — a static site with no build step. Computation runs locally in the browser.
| Feature | What it does |
|---|---|
| Profiles | Switch between the KZ EDX Pro and Redmi 12 speaker chains. |
| Response chart | Displays response derived from the hosted IRS, with log-frequency scale and 1/6-octave smoothing. |
| Quality assurance | Fetches and checks hosted VDC, IRS, and EEL assets in the browser. Downloads are blocked when QA fails. |
| Export | Downloads the three profile files or copies a 127-point GraphicEQ curve derived from the IRS. |
| Verification harness | Runs offline checks. This is not native app, ARM, or device validation. |
| Documentation reader | Opens repository documents in the site. Example deep link: #read/docs/methodology.md. |
| Reference | Purpose | Read |
|---|---|---|
| VDC reference | Shipped VDC structure and evidence boundaries | Site |
| IRS functions | IRS format, metrics, and validation | Site |
| EEL2 functions | VM syntax, parser constraints, and implementation guidance | Site |
| System reference | Current architecture, profile state, operation, evidence, and limitations | Site |
| Methodology | Synthesis, gating, and evidence method | Site |
| Test report | Results, scope, and open questions | Site |
| Speaker setup | Redmi 12 installation instructions | Site |
| IEM setup | KZ EDX Pro installation instructions | Site |
| Speaker report | Speaker-chain numerical report | Site |
| IEM report | IEM-chain numerical report | Site |
| Attribution | Third-party sources and credits | Site |
| License notes | Project license scope and notices | GitHub |
| Component | Offline / numerical | App or device | Listening |
|---|---|---|---|
| Speaker VDC | PASS · identity | USER REPORTED | USER REPORTED |
| Speaker IRS | PASS | USER REPORTED | USER REPORTED |
| Speaker EEL | PARTIAL · compile, ceiling, bypass; non-finite rerun pending | USER REPORTED | USER REPORTED |
| IEM VDC | PASS · identity | — | — |
| IEM IRS | PASS | NOT RUN | NOT RUN |
| IEM EEL effects path | Prior v4 checks: effects, ceiling, bypass, non-finite | NOT RUN for current v5 | NOT RUN |
| IEM EEL v5 mode selector | Source review only; VM retest pending | NOT RUN | NOT RUN |
Evidence boundary: Offline checks use a source-built JamesDSP VM on x86_64. They are not native ARM, RootlessJamesDSP app, target-device, or listening tests. “Numerically qualified” does not imply device validation. Full scope and outstanding work: test report.
- Reproducibility:
iem_build.py,generate_irs.py, and complete source inputs are not included. The published IRS cannot be regenerated from this repository alone. - Speaker IRS provenance: The measurement setup is not documented here. If a microphone response was included in the source measurement, it is also present in the resulting filter.
- IEM voicing: The IEM voicing has no measurement behind it; it is explicitly a design choice.
- Spatial processing: The IEM surround feature is crossfeed plus width, not HRTF virtualisation.
- ARM compatibility: EEL tests were run on x86_64. The v5 processing-mode selector has not yet been VM- or device-tested.
- Limiter scope: Sample-peak limiting is not true-peak limiting or driver-excursion protection.
- Sample rate: The IRS assets are 48 kHz only. Using another host rate changes the effective impulse-response timing and frequency behaviour.
Verify downloaded files with sha256sum.
SHA-256 checksums
Redmi_12/HA_Speaker_Master.vdc
933be0b02daead88450f2042065d0fb5a44e77099f8c29a6e37b9669e6a2d6ef
Redmi_12/HA_Speaker_Master.irs
0ba8da0f4351aca761e3c7e39b37bccf9b61403f7d92ed60ecf1b23cda0c7de4
Redmi_12/HA_Speaker_Master.eel
e5cfbc49cb82aad8d3c45da5ca27715a9712639cb1f41aa3dce2a9677c903af4
edx_pro/HA_IEM_v3.vdc
933be0b02daead88450f2042065d0fb5a44e77099f8c29a6e37b9669e6a2d6ef
edx_pro/HA_IEM_v3.irs
b6d751a1613b48569a6ec485f477603172dcc14b0072f0c86c93b4668363b9d5
edx_pro/HA_IEM_v3.eel
3efa4976c3e9788faafa0225a9bcffa5944c04cb3d1bbfd5b64ce3a6d7d56b17
Show repository tree
HyperAudioo/
├── README.md
├── index.html
├── ATTRIBUTION.md
├── LICENSE
├── LICENSE-MIT
├── LICENSE-CC-BY-SA-4.0
├── LICENSE.md
├── Redmi_12/
│ ├── HA_Speaker_Master.vdc
│ ├── HA_Speaker_Master.irs
│ ├── HA_Speaker_Master.eel
│ ├── HA_Speaker_Master_BETA.eel
│ ├── Setting_Guide_S.md
│ └── report_Redmi_12.md
├── edx_pro/
│ ├── HA_IEM_v3.vdc
│ ├── HA_IEM_v3.irs
│ ├── HA_IEM_v3.eel
│ ├── Setting_Guide_IEM.md
│ └── report_EDX_HA.md
├── docs/
│ ├── vdc_functions.md
│ ├── irs_functions.md
│ ├── eel_functions.md
│ ├── system_full.md
│ ├── methodology.md
│ └── test_report.md
├── assets/
│ ├── logo.svg
│ ├── banner.svg
│ ├── style.css
│ ├── script.js
│ ├── docs.js
│ ├── fonts/
│ └── verification harness graphics
└── .github/workflows/
└── ci.yml
| Scope | License |
|---|---|
EEL2 code (.eel) |
MIT |
Audio artifacts (.irs, .vdc) and reports |
CC BY-SA 4.0 |
See ATTRIBUTION.md. Presets derive from AutoEq-format data (jaakkopasanen/AutoEq, MIT) via RikudouGoku's database, corroborated by Speaker Ranking. EEL2 scripts target the VM in RootlessJamesDSP (ThePBone, Apache 2.0). No third-party code or raw measurements are redistributed.
Designed & tuned by Noushad M. · © 2026