feat: inline timestamp markers for offline audio segments

[mdmohsin7]

Summary

• Embeds inline 0xFF timestamp markers in the offline audio data stream so each recording segment (between BLE disconnects or power cycles) has an exact UTC start time
• Extends info.txt on SD card from 4 to 8 bytes to persist a recording start time alongside the read offset
• Exposes recording start time via the BLE storage characteristic (3rd uint32 value)
• Fixes WAL timing bugs: chunk size now uses actual codec FPS, and WAL duration derives from real frame count instead of a hardcoded 60-second assumption
• Flushes the storage write buffer on device shutdown to prevent data loss

Firmware changes

• transport.c: write_timestamp_to_storage() injects a 5-byte marker (0xFF + 4-byte LE epoch) before the first offline audio frame of each segment. storage_flush_buffer() flushes pending data on shutdown. needs_timestamp_marker flag resets on each BLE connect.
• sd_card.c: recording_start_time persisted in info.txt (backward-compatible 8-byte format)
• storage.c: BLE read characteristic returns [file_size, offset, recording_start_time]
• button.c: Calls storage_flush_buffer() before app_sd_off() in turnoff_all()

App changes

• sdcard_wal_sync.dart: Parses 0xFF markers in both BLE and WiFi sync paths, splits audio into correctly-timed WAL segments. Fixes chunk size from sdcardChunkSizeSecs * 100 to sdcardChunkSizeSecs * codec.getFramesPerSecond().

Backward compatibility

• New firmware + old app: Old parser hits bounds check on 255-byte frame size and skips to next block. Minimal data loss (5 bytes per marker).
• Old firmware + new app: No markers in stream, new code path never triggers. Works unchanged.
• info.txt: Old 4-byte format reads correctly (extra bytes default to zero).

Test plan

• Build firmware via Docker and flash OTA
• Connect phone → disconnect BLE → record ~1 min offline → reconnect briefly → disconnect → record ~1 min → sync
• Verify app shows two WAL segments with correct timestamps and a gap between them
• Test power off during offline recording → power on → record → sync → verify two segments
• Test with old app version to confirm backward compatibility

[mdmohsin7]

[gemini-code-assist]

Code Review

This pull request successfully implements a valuable feature by embedding inline timestamp markers in offline audio recordings, allowing for accurate segmentation. The firmware and app changes are well-executed, with robust logic for handling the new data format and maintaining backward compatibility. However, the review identified two high-severity issues that should be addressed: unrelated logging configuration changes in omi.conf that appear to be development artifacts, and the removal of a necessary header file in sd_card.c which could compromise build stability. Once these points are resolved, the pull request will be in excellent shape.

[mdmohsin7]

Split into two separate PRs: #5571 (firmware) and #5572 (app)

[github-actions]

Hey @mdmohsin7 👋

Thank you so much for taking the time to contribute to Omi! We truly appreciate you putting in the effort to submit this pull request.

After careful review, we've decided not to merge this particular PR. Please don't take this personally — we genuinely try to merge as many contributions as possible, but sometimes we have to make tough calls based on:

• Project standards — Ensuring consistency across the codebase
• User needs — Making sure changes align with what our users need
• Code best practices — Maintaining code quality and maintainability
• Project direction — Keeping aligned with our roadmap and vision

Your contribution is still valuable to us, and we'd love to see you contribute again in the future! If you'd like feedback on how to improve this PR or want to discuss alternative approaches, please don't hesitate to reach out.

Thank you for being part of the Omi community! 💜