Scriber Optics (Gen 2) — v127 regression: audio muted every 1.35 s in captured video
My Ray-Ban Meta Scriber Optics (Gen 2) developed a periodic audio dropout in locally captured video immediately after the v127 firmware update. Posting the full analysis in case anyone else on v127 can reproduce it.
WHAT HAPPENS
Every 1.352355 seconds - exactly 64,913 samples at 48 kHz - the audio track produces one full-scale transient (-0.4 dBFS, about +59 dB above the median level) followed by 130-400 ms of hard mute. Total audio loss is 13-28% of every recording. It sounds like the volume is cycling loud and quiet about once a second.
The muted segments contain exact digital zeros, with RMS down to -180 dBFS. A microphone or acoustic fault cannot produce digital silence, so this is a software mute.
IT IS A REGRESSION
Recordings from Aug 2 and Aug 9: zero dropouts, zero samples above 0.9 full scale, no periodicity at all (autocorrelation at 1.352 s is r = 0.03 and r = -0.05). Every recording from Aug 11 onward shows the fault (autocorrelation r = 0.74). The v127 update was installed between Aug 9 and Aug 11.
WHAT I RULED OUT
The fault period is identical across all affected clips to within 0.0002%, and is completely unaffected by resolution (1080p/30 vs 3K/30), video bitrate (14.9 vs 24.7 Mbps), or stabilisation (medium vs none).
The Bluetooth headset path (HFP, recorded on the phone with the glasses as the microphone) is NOT affected. Only the on-glasses capture pipeline is broken, which points at the audio capture clock or buffer configuration rather than the microphone hardware.
One detail that may help whoever looks at this: 65536 / 64913 = 1.0096, consistent with a roughly 0.96% sample-rate mismatch forcing a resync every 2^16 samples.
ALREADY TRIED, NONE OF IT HELPED
Factory reset. Unpair and re-pair on both iOS and Android. Every combination of video resolution, duration and stabilisation settings. Turning off Hey Meta, wake word chime, direct reply and in-call voice commands - this made recordings worse, not better.
SETUP
Ray-Ban Meta Scriber Optics (Gen 2), firmware v127, purchased in the US, paired to an iPhone.
Attached is a waveform analysis of five recordings, two from before the update and three from after. I still have all the original video files and can share them with anyone from Meta who wants them.
If you are on v127 with any Gen 2 model, could you record a 30-second video and check whether the audio cuts out about once a second? I would like to know whether this is specific to Scriber Optics or affects the whole Gen 2 line.