fix(stt): chunked transcription with progress, retry, and no 300s cliff by EtienneLescot · Pull Request #232 · getopenscreen/openscreen

fix(stt): transcribe in chunks, with progress and per-chunk retry

EtienneLescot

committed

d108476

feat(stt): show real transcription progress instead of an endless spinner

EtienneLescot

committed

9417185

fix(stt): type the caption status mock for the progress fields

EtienneLescot

committed

afec7bc

fix(stt): make the chunked run cancellable, and its failures legible

EtienneLescot

committed

4a9a7be

fix(ui): let the progress bar own its spacing, and name the model download

EtienneLescot

committed

2bb4e34

fix(build): bound a stalled ffmpeg fetch, and read the nix version

EtienneLescot

committed

2578a98

fix(ui): show the detected language where it can actually be seen

EtienneLescot

committed

bea568e

perf(transcript): stop re-rendering every word to move one underline

EtienneLescot

committed

ccbe526

feat(media): strip the Media timeline down to arranging

EtienneLescot

committed

a29b393

perf(waveform): share the decode; stop holding 714 MB to draw 400 bars

EtienneLescot

committed

490e684

perf(waveform): decode peaks natively, once per recording ever

EtienneLescot

committed

1a53c52

Merge branch 'release/v1.8.0' into fix/stt-chunked-transcription

EtienneLescot

authored

83b1157

Navigation Menu

Repository navigation

fix(stt): chunked transcription with progress, retry, and no 300s cliff - #232

fix(stt): chunked transcription with progress, retry, and no 300s cliff#232
EtienneLescot merged 12 commits into
release/v1.8.0from
fix/stt-chunked-transcription

Commits on Aug 3, 2026

fix(stt): transcribe in chunks, with progress and per-chunk retry

feat(stt): show real transcription progress instead of an endless spinner

fix(stt): type the caption status mock for the progress fields

fix(stt): make the chunked run cancellable, and its failures legible

fix(ui): let the progress bar own its spacing, and name the model download

fix(build): bound a stalled ffmpeg fetch, and read the nix version

fix(ui): show the detected language where it can actually be seen

perf(transcript): stop re-rendering every word to move one underline

feat(media): strip the Media timeline down to arranging

perf(waveform): share the decode; stop holding 714 MB to draw 400 bars

perf(waveform): decode peaks natively, once per recording ever

Merge branch 'release/v1.8.0' into fix/stt-chunked-transcription