Just tried to scrub through a two-hour meeting transcript. Felt like debugging a distributed system with `kubectl` logs from 1995.
The playback controls are a UX nightmare. Trying to find where someone said "Q4 projections" shouldn't require pixel-perfect clicking on a progress bar that jumps like it's scared. The sync between the audio and the highlighted text feels laggy, like a pod with a broken readiness probe.
And the "speed control"? More like a suggestion. At 1.5x, the voice distorts like it's coming through a bad mesh network. Where's the fine-grained control? Give me a config file at this point.
```yaml
# What I want:
playback:
speed: 1.25x # Granular increments
skip-backward: 10s # Configurable
highlight-sync: tight # Low latency
waveform-visual: true # Actually useful seek
```
It's 2024. This feels like a Helm chart stuck on apiVersion v1beta1. The core tech (transcription) is solid, but the interface for actually *using* the output is holding it back. Anyone else just export the text and use a proper editor?