Included features
- ADM format, metadata, loudness and true-peak checks with reports.
- Ship delivery plans, file copies and SHA-256 manifests.
- Encode comparison reports are advisory; platform acceptance is not guaranteed.
QC
Check delivery specs before the rejection email.
Preflight an Atmos ADM BWF master, read each Pass, Warn, Fail or Skip, and inspect the report before delivery.
Twenty-two checks across Apple Music, Amazon, TIDAL, UMG, Sony. Read the rejection cause before the rejection letter arrives.
macOS 13.0 or later · Apple Silicon and Intel · Developer ID signed · Apple notarized · 30-day trial
Version 1.0.10
What is included
UTUVO QC · US$19 · One app, one-time purchase. Other UTUVO apps are separate purchases.
| System requirements | macOS 13.0 or later · Apple Silicon and Intel |
|---|---|
| Licence | No subscription. Buy once, use the purchased major version and every update within that major. 3 of your own Macs. Terms |
| Move or reinstall | Open the app on the old Mac, go to Settings and deactivate the licence there, then enter the same key on the new Mac. No email needed, and a licence covers 3 of your own Macs. Deactivate while the old Mac still has a network connection: if it is offline at that moment the app clears the licence locally but the seat can stay held on our side, and support@utuvo.app can release it for you. For reinstallation, use the same key. If activation cannot be completed, contact support with your order details. |
| Trial | Download and open the app for a 30-day full-feature trial. No card or account required. No automatic charge. |
| Refunds and support | 14-calendar-day no-reason refund period. Refund policy · support@utuvo.app Include your order number for refunds, or app/macOS versions and reproduction steps for issues. |
| Updates | Read the update and maintenance policy ↗ |
Buy for 3 Macs (US$9.50)LAUNCH50: 50% off, US$9.50 per Mac app. Applied at checkout through Nov 30, 2026, 23:59 (Taipei time).
Product demo
See QC find the problem before delivery.
Twenty-two delivery checks, one clear verdict per row, with the reason behind every warning or failure.
The rejection reasons are public. Nobody checks them all at once before you upload.
Platforms publish the rules. They rarely tell you which one you broke. Below are eight rejection causes the delivery specs name, each paired with the QC check that catches it, and the full table of all twenty-two checks against the threshold and confidence that drives the verdict.
Eight rejection causes. One named check each.
- Sample rate not 48 kHz. Caught by: Sample rate check.
- Bit depth under 24-bit. Caught by: Bit depth check.
- ADM metadata missing chna or axml. Caught by: ADM metadata check.
- No bed at all in the file. Caught by: Bed layout check.
- LFE missing, doubled, or full-band. Caught by: LFE channel check + LFE band-limit check.
- True peak above −1 dBTP. Caught by: True peak check (4x oversampling, BS.1770-4).
- Timeline offset left in the export. Caught by: Programme start check.
- Timecode frame rate not 24 fps. Caught by: Timecode frame rate check (Apple / UMG / Sony: Confirmed MUST).
These checks target the rejection reasons the specs actually name. Every result is a pre-delivery reference, not the sole basis for whether a platform accepts the file; the honest scope is that the known rejection causes have been checked.
All twenty-two checks, with thresholds and confidence.
Confirmed is a hard must against the spec the platform publishes. Advisory is a recommendation; the platform may still accept, but the engineer should know.
| Check | Threshold | Confidence |
|---|---|---|
| Sample rate | 48 kHz | Confirmed |
| Bit depth | ≥ 24-bit | Confirmed |
| ADM metadata | chna + axml present | Confirmed |
| Bed layout | 7.1.2 baseline | Advisory |
| LFE channel | exactly one | Confirmed |
| LFE band-limit | LF only (HF ratio) | Advisory |
| Object count | ≤ 118 | Confirmed |
| Object positions | ±1.0, finite, every dynamic keyframe | Confirmed |
| Programme start | 00:00:00.00000 | Advisory |
| Binaural metadata | not all Off | Advisory |
| Channel count | ≤ 128 | Advisory |
| Silent objects | none silent | Advisory |
| Height content | real height present | Advisory |
| True peak | ≤ −1 dBTP (BS.1770-4, 4x oversampling) | Confirmed |
| Integrated loudness | −20 to −18 LUFS (Apple window) | Advisory |
| Stereo alignment | ≤ 50 ms | Advisory |
| Duration match | ≤ 41.7 ms | Advisory |
| Dialnorm | present and consistent | Confirmed |
| Spatialization authenticity | native vs upmix score | Advisory |
| File naming | platform naming rules | Advisory |
| DAMF package | .atmos attached at 96 kHz | Confirmed |
| Timecode frame rate | 24 fps | Confirmed |
Your rejection letter is hidden in metadata. QC pulls the whole dbmd apart and lays it out.
The Dolby metadata chunk (dbmd) carries dialnorm, downmix coefficients, per-layout monitoring trims and the timecode frame rate. Most rejection letters never mention it, because most readers never open it. QC parses dbmd and the .atmos companion so the engineer sees what the delivery spec actually asks for.
dialnorm, downmix, and the per-layout trims.
- Dialnorm and downmix levels: parsed from the dbmd chunk (EBU Tech 3285 section 6), shown alongside the spec the platform expects.
- Per-layout monitoring trims: the Dolby-private dbmd seg10 holds the trim for each monitoring layout, either a Manual number or the marker automatic when the renderer computes it on the fly.
- Timecode frame rate: read from both dbmd and the .atmos companion file. Apple Music, UMG and Sony all mark 24 fps as Confirmed MUST.
Others give you one number. We tell you how far that number is from Dolby Renderer, and why.
Since 1.0.10, UTUVO QC measures loudness on a Renderer-style 5.1 loudness render of the ADM, with the master’s own render mode and trims (ITU-R BS.1770-4). On 396 test masters we built ourselves, it matches Dolby Atmos Renderer 5.5 on Integrated for 380 of 384, True Peak for 339 of 396 and Loudness Range for all 384. Every remaining difference is one 0.1 display step, on a value within 0.04 dB of a rounding edge.
Honest by construction.
- Measurement-only loudness render. QC does not produce a delivery render or an encode, and never modifies your master.
- No Dolby SDK, no licence. QC reads ADM metadata directly and measures loudness from the PCM.
- Fully offline. No telemetry, no upload, no cloud. Masters and reports stay on your machine.
Every result names its measurement basis. Size or diffuse objects, non-neutral balances and other cases outside the validated profiles fall back to an internal 7.1.4 render, labelled as not calibrated to Dolby Renderer. QC is not a certified Dolby-equivalent meter. When a delivery must quote the Renderer’s own figures, import its Entire-master Summary report: QC checks it belongs to the same master, then shows Dolby’s values.
Fake Atmos can be heard. Now it can be seen.
Spatialization authenticity scores how likely a master is a native discrete Atmos mix rather than a stereo-to-Atmos upmix. ObjectMap adds a top-down view and a Height view, and draws each dynamic object’s trajectory as a flight trace through the room.
Authenticity: experimental, advisory, opt-in.
- Spatialization authenticity is experimental, advisory, and opt-in. It can only ever return Warn. It is never a verdict.
- ObjectMap shows top-down and Height views with dynamic flight traces per object, read from the ADM at every dynamic keyframe.
- Object motion summary: static, horizontal or with-height, read from the metadata.
Upmixing a stereo catalogue to Atmos is a legitimate practice; upmixed masters routinely pass platform QC and ship. This check answers "native discrete or upmixed?", not "valid or invalid?".
A folder at a time. Same engine in GUI and CLI.
Per-file reports, worst-case exit code.
Drop a folder, get one HTML report per master, and the batch ends with exit 0 (all pass), 1 (any warn) or 2 (any fail). GUI and CLI run the same Rust engine, so a delivery script on the studio NAS and a drag-and-drop session in the GUI share the same verdict.
- HTML report output, self-contained, prints to PDF without external services.
- Same engine in CLI and GUI. Ship one verdict, in a script or a window.
- Three-language manual bundled: English, Traditional Chinese, Simplified Chinese.
MP4: container-level checks, no faking.
Atmos MP4 (DD+ JOC) gets format and channel info read from the container. That is the honest scope: container-level checks only, not a full Atmos MP4 QC. When the input is an MP4, the report shows the red "Not a Deliverable Master" stamp so the engineer is never misled about what was checked.
Passed QC. Now ship it the way the label wants it.
The Ship tab takes the master that just passed and builds the delivery folder for you: tree, file names, manifest, checksums. Nothing is written until you have read the plan.
Five profiles, one dry-run first.
- Sony Music Atmos ADM BWF + DAMF, Dolby generic ADM BWF, UMG Dolby Atmos Audio Servicing, UMG archival, and a distributor-generic profile. UMG profiles ask for mixer initials, studio location, delivery date and resubmission status.
- QC stays the gate: a Fail refuses to package, a Skip can only dry-run, the stereo reference is the one you set on the QC tab.
- Preview delivery lists every file with its destination and size. Anything that would stop the delivery shows in the Blocked card; anything worth reading first shows in Warnings.
- Sources open read-only and copy byte for byte. The folder ships with manifest.json, SHA256SUMS and a delivery note; the recipient checks it with utuvo-qc verify, no GUI needed.
After the platform encodes it, how much is left?
The Encode tab plays the EC-3 JOC through a 7.1.4 endpoint you choose, captures the system render in real time, aligns it to the ADM master and measures the difference channel by channel.
Measurements, not a verdict.
- Per-channel lag and gain, integrated loudness and true peak delta, residual energy, and residual by band: below 120 Hz, 120 Hz to 2 kHz, 2 to 8 kHz, above 8 kHz.
- Encoding is lossy, so the residual is never zero. Encode reports and advises; it never adds a Fail to the QC verdict.
- Capture runs at playback speed: a seven-minute deliverable takes seven minutes. The tab shows progress and the remaining estimate.
- Same engine in the CLI: utuvo-qc encode-check writes JSON and an HTML section that drops into the QC report.
Learn it in the app.
A guided tour walks the QC, Encode and Ship tabs with a spotlight and a pointer. Steps that need an action wait for you; steps that only explain move on with Got it. Replay it from the ? button in the header, from Settings, or from the Help menu.
Requirements
macOS 13.0 or later · Apple Silicon and Intel
Status: Available. Download 1.0.10 above.
Specs and download
| Version | 1.0.10 |
|---|---|
| Requirements | macOS 13.0 or later · Apple Silicon and Intel |
| Size | 19,523,551 bytes |
| SHA-256 | c54a9251dc28367bb5d5fecda7d1f29325fe5c6475e7b71a4a1d1d607a415776 |
| Signing | Developer ID signed · Apple notarized |
| Trial | 30 days, then a licence key |
Fewer checks to juggle. A clearer workflow.
A comparison of manual, separate steps and this tool’s workflow. Other tools can also do these jobs.
| Without UTUVO | With UTUVO |
|---|---|
| Read ADM metadata separately from the delivery checklist. | Inspect ADM metadata against the selected delivery profile. |
| Assemble loudness results from a separate measurement workflow. | Read the supported loudness measurements in the QC report. |
| Track unresolved checks in notes. | See each check as Pass, Warn, Fail or Skip. |
| Prepare a report for each master by hand. | Run folder batches and export self-contained HTML reports. |
| Copy passed masters and build a checksum list manually. | Review the Ship dry-run, then create copies with a SHA-256 manifest. |
Asked often enough.
Do preflight checks change my source master?
Preflight reads the ADM master and reports its checks. It does not repair the source master. QC also has separate Encode and Ship workflows; the report is not a guarantee that a platform will accept a delivery.
Can I put it in a delivery script?
Yes. CLI with automation exit codes (0/1/2; batch takes the worst result), plus folder batch checking and a self-contained HTML report you can print to PDF.
Does anything leave my machine?
No. Fully offline, zero telemetry. Masters and reports stay local.
How does the 30-day trial work without a card?
Download and open the app to evaluate all features for 30 days. No credit card or account is required. The trial does not turn into an automatic charge; purchasing is a separate decision.
Does a perpetual licence include every future version?
It includes the major version you purchase and every update released within that major version. A future major version may be a separate optional one-time purchase or may be free. It is never charged automatically. If you buy within 30 calendar days before a new major becomes available, that new major is included at no charge.
I replaced my Mac. How do I move my licence?
Open the app on the old Mac, go to Settings and deactivate the licence there, then enter the same key on the new Mac. No email needed, and a licence covers 3 of your own Macs. Deactivate while the old Mac still has a network connection: if it is offline at that moment the app clears the licence locally but the seat can stay held on our side, and support@utuvo.app can release it for you.
Which macOS version is required? Does it run on Intel?
The current download requires macOS 13.0 or later · Apple Silicon and Intel. Apple Silicon and Intel are supported; no Windows build. Try it on your working Mac before deciding to buy.
How do the 30-day trial and 14-day refund period differ?
The 30-day full-feature trial is for evaluation before purchase. The 14-calendar-day no-reason refund period begins separately on the purchase date, so trial days do not reduce it. After 14 days, a simple change of mind is generally not refundable, while transaction errors, activation failure, nonconformity and statutory remedies remain covered.
How do I request a refund or get support?
Email support@utuvo.app. Within 14 days of purchase you can request a full refund without giving a reason; include your order number or licence key, or use the link in your order confirmation email. For a product issue, include the app version, macOS version and steps to reproduce it.
Will offline use or a verification outage stop my paid app?
No. After a paid licence is successfully activated, it continues to work offline on that device. A network or verification-service failure alone cannot deactivate it. If the verification service permanently ends, we will provide a free way for lawful perpetual-licence holders to continue using, reinstalling and activating on a replacement compatible Mac within their device count.
UTUVO Work Series
Part of the UTUVO Work Series.
Five independent tools for five different jobs around a session.
UTUVO ScanMissing is a lazy answer.Available for macOS
UTUVO PrepNames and heads. Two decisions, then it is done.Available for macOS
UTUVO QCEvery known rejection reason, caught before you upload.Available for macOS
UTUVO ManagerKnow your plugin rig before the session finds the problem.Available for macOS
UTUVO CatalogA library that respects the masters.Invitation-only beta
These tools come from someone who uses them to deliver every day.
UTUVO is built by one person: Micky Yang.
He's a Tonmeister with 1,600+ Mandopop songs mixed in Dolby Atmos, four Golden Melody Awards for recording, and he teaches recording at university. Most tools are what engineers imagine recordists need. UTUVO is what a recordist built because he couldn't stand the existing ones.
- 4× Golden Melody Awards (Taiwan's Grammys) for recording
- Apple Music made him an official engineer playlist: "Micky Yang: The Engineers"
- Game scores: NetEase's Onmyoji, Harry Potter: Magic Awakened
- 1,000+ Chinese pop string sessions