Recorded engine builds
Distinct native binary fingerprints in the Studio release history through October 30.
← FNT Software SuiteGood sound takes
19 52 6h 53m 670
THE ENGINE / EVIDENCE SNAPSHOT / 30 OCT 2026
Good sound takes
more than an idea.
The original FNT audio engine is being built through repeated implementation, captured trials and release validation. These are the figures the records support.
Detailed bench trials
Distinct hardware trial reports in the testing history compiled October 27.
Captured bench input
Cumulative recorded input across those trials, including nine completed 30-minute runs.
Release checks passed
603 application + 67 desktop checks recorded for Studio 1.12.0 / Engine 1.6.0.
WHAT WE COUNTED
Numbers with
something behind them.
These measurements cover different scopes. They are presented separately so the history stays meaningful.
Distinct native engine binaries
Counted by unique, non-empty nativeEngine.sha256 values in the tracked Studio installer feed. The audit found 54 fingerprint-bearing feed records covering 50 application versions and 19 different executable hashes. An application update can reuse the same engine.
Source: Studio dist/central-installer.json history through commit 9062a6d, October 30, 2026. Three early feed records have no fingerprint. This is a conservative recorded-build count, not every experimental compile.
Detailed historical hardware trial reports
The October 27 bench history reconstructs distinct trials T01–T52 at 48 kHz, merging two duplicate uploads. A summary-only startup failure is excluded from both the detailed-report count and capture duration.
The report covers 14 documented bench iterations, 0.1.0–0.2.3. Those iterations are a separate lineage and are not added to the 19 native binaries.
Source: FNT Audio Bench Test History, compiled October 27, 2026, pages 1–3, 5 and 10–13.
Cumulative captured input
Total captured input across the detailed historical bench trials. Nine trials reached a full 30 minutes. This is accumulated capture time, not one continuous soak or a measure of development hours.
Trial completion could include timing warnings. The documented 64-sample Mixed68 endurance attempt remained unresolved; current Windows/ASIO hardware qualification is still open.
Source: the same October 27 reconstructed hardware trial ledger.
Recorded Studio 1.12.0 release checks
603 application checks and 67 desktop checks passed, with zero recorded failures. The current Studio 1.12.1 capture uses the same Engine 1.6.0 binary. Another 68 Connect updater checks are reported separately.
The focused 21-check rerun overlaps the full suite and is excluded. These are per-release application and desktop checks, not 670 engine-only tests or a deduplicated lifetime total.
Sources: docs/release-1.12.0-validation.json, docs/precision-studio-1.12.0.md and docs/precision-engine-build.json.
ONE SCOPED PERFORMANCE CASE STUDY
Measure the path.
Show the limits.
Fixed portable fixtures compared against the same baseline. These timings describe specific code paths on the report host.
TRIM / PAN MIXER FIXTURE
28.5%Less CPU time in this fixture
Median of five runs per build. 32 monitored stereo channels, 48 kHz, no inserts, +0.7 dB Pre and varied pan, processing one second of deterministic stereo input.
EQ / FET / AMP-HEAVY FIXTURE
Output matched.
No speed gain.
The companion fixture was 1.3% slower: 3,908.08 ms baseline versus 3,957.42 ms candidate, median of three runs each. Its one-second outputs matched byte for byte.
We do not generalize the mixer result to total engine speed, real-time capacity, device latency or dropouts.
Baseline source commit d8a19e0 · Candidate: Studio 1.12.0 / Engine 1.6.0 · Portable C++, g++ -std=c++17 -O2 · Sources: precision-mixer-bench.json and precision-bench-results.json.
THE NEXT MEASUREMENT MATTERS
Built to improve.
Honest about what’s left.
Portable processing fixtures, browser checks and Windows cross-compilation provide useful coverage. They do not establish current physical ASIO performance across interfaces, sustained low-buffer endurance, licensed commercial plug-in compatibility or listening equivalence to original hardware.
The next priorities are physical device qualification, repeatable long sessions and practical compatibility testing. The history includes failed trials because those results shape the next build too.