[🐞 BUG] Models missing big sections of speech

[caddisoncallaghan]

Describe the bug
In all models, I appear to be missing whole and partial sentences. Running in Mac OS Voice memo has no such issues.

To Reproduce
Steps to reproduce the behavior:

1. Go to file transcription
2. upload .m4a file
3. transcribe with model of choice (used Parakeet v2 &v3, Apple Mac OS 26+, Apple Legacy
4. Review transcribed text and compare with audio recording and text output from Voice Memos App

Expected behavior
I would expect transcriptions that match/exceed that of voice memos app

Screenshots
If applicable, add screenshots.

Environment:

• macOS Version: 26.0.1
• App Version: 1.5.10
• Architecture: M1

Additional context
Add any other context about the problem here.

Crash Logs
If the app crashed, please attach relevant crash logs

[altic-dev]

Can you give me audio clips that you think doesn't work?