Repository navigation
iOS 1.0.2
This release updates the final audio stage while keeping one stereo output per platform.
- Apply the documented
rawprofile at startup; it previously retained the conservative source gain. - Remove the mandatory 10 ms final-output lookahead with shared stereo sample-peak protection.
- Use the shared linear 10 ms master ramp, including volume changes published during a callback.
- Exercise production rendering and startup profile selection in host regressions.
- Add Apple preset calibration across six voice counts, with stress results reported separately.
The final stage retains zero samples. Overload changes the waveform, and gain recovers with a 100 ms exponential time constant. Sample ceilings remain 0.85, or 0.70 for speakerSafe. Public methods, fixed event gains and chord compensation remain unchanged.
Apps that request raw now receive its documented higher source gain. Review their gain staging when updating.
Production-renderer checks cover 16, 44.1, 48, 96 and 192 kHz. The matched host comparison preserved all 40 cases without overload within an absolute float tolerance of 1e-7 after aligning the old delay. Studio and Frontier comparisons ran on an Android device and an iPad; route-wide latency, Bluetooth behavior and extended maximum-load acceptance remain outside those results.
The public JavaScript API remains at version 1.0.0. Requires Titanium SDK 13.2.0.GA or newer and iOS 15 or newer. Device, simulator and Mac Catalyst builds are included.
Install the attached ti.synthengine-iphone-1.0.2.zip. The other package is in Android 1.0.2.
Built from 4a45f21. Linux and macOS CI passed. See validation and compatibility for scope and remaining limits.