Oyma streams audio buffers directly to local temporary disk storage in real time. Even if your Mac experiences severe RAM pressure or crashes, recorded audio is never lost and recovers automatically upon restart.
Resilient Disk-Buffered Architecture
A critical vulnerability in poorly designed recording tools is holding multi-hour audio streams entirely in volatile RAM. If the operating system runs low on memory, or if an unexpected crash occurs, the in-memory buffer vanishes, destroying hours of critical client discussions.
Oyma is engineered with high-reliability disk streaming invariants.
How Oyma Prevents Data Loss
- Continuous Chunk Flushing: Captured audio frames from ScreenCaptureKit and CoreAudio are committed to an unbuffered temporary disk file every 5 seconds.
- Minimal RAM Footprint: The active recording engine consumes less than 85 MB of system RAM regardless of whether your meeting lasts 15 minutes or 6 hours.
- Crash Recovery Ledger: If your MacBook runs out of battery, experiences an unexpected kernel panic, or force-closes:
- Upon the next application launch, Oyma detects the unfinalized audio file.
- The application automatically repairs the container headers (
moovatom in.m4afiles) using native AVFoundation recovery tools. - The repaired audio is preserved in your vault and transcription resumes immediately.
[Audio Frames] -> [Ring Buffer RAM (5 sec)] -> [Continuous Disk Write (NVMe)]
|
v
[Automatic Crash Recovery]
Memory Pressure Mitigation on 8 GB Macs
On entry-level Macs with 8 GB of unified memory, running heavy video calls alongside Chrome can trigger macOS memory compression. Oyma monitors system memory pressure warnings (DISPATCH_SOURCE_TYPE_MEMORYPRESSURE) and dynamically scales down UI animations and background indexing to prioritize audio recording integrity above all else.
Memory Management and OS Safety Measures
On systems operating near physical memory capacity:
- macOS Memory Pressure Monitoring: When macOS signals high memory usage, Oyma reduces in-memory thumbnail caches and suspends background search re-indexing.
- Disk-First Audio Buffering: Recorded audio frames bypass RAM caches and write directly to temporary NVMe disk storage via kernel file descriptors.
- Safe State Recovery: Even if the Mac experiences an emergency kernel reboot, the partially written recording remains intact on disk, ready for automatic repair upon relaunch.
Resilient Disk Buffering and Container Repair
Audio is written to disk in continuous 5-second unbuffered chunks. Even in the event of an unexpected operating system reboot or sudden battery exhaustion, previously recorded audio frames remain completely safe and recoverable on disk.
Upon relaunch, Oyma‘s recovery engine scans for unfinalized .m4a files, reconstructs missing container metadata headers, and opens the repaired recording in your vault ready for transcription.
Memory Safety and Resilience Specifications
- Disk Streaming: Flushes audio frames to disk every 5 seconds via unbuffered file descriptors.
- RAM Footprint: Active recording engine consumes under 85 MB of memory regardless of call duration.
- Container Repair: Automatically reconstructs MP4/M4A metadata headers upon restart following sudden power loss.
- APFS Safety: Copy-on-write filesystem semantics prevent corruption of existing vault files.