🎉 Vocalinux v0.11.0-beta
24 commits merged since v0.10.2-beta — new Advanced Settings tab, IBus/recognition hardening, and broad distro compatibility improvements.
✨ New Features
Advanced Settings Tab (#415)
A new Advanced tab in the Settings dialog exposes whisper.cpp anti-hallucination parameters directly from the UI — no config file editing required:
- Temperature — control inference randomness (lower = more deterministic)
- No-speech threshold — tune sensitivity to background noise
- Max segment length — limit how long transcribed chunks can be
- Additional inference-time parameters for fine-grained control
Screencast.From.2026-05-06.22-35-14.webm
This addresses long-standing requests from @ericzundel (#363), @christophtschmidt (#247), and @bradwood (#393) for more control over hallucination and noise transcription behaviour.
🐛 Bug Fixes
🎤 Speech Recognition
- Preserve final speech on stop — last spoken words no longer truncated when stopping recognition (reported by @sabbari in #401)
- Stop sound plays immediately on shortcut release — no more delayed audio feedback (#426, by @juanfradb)
- Stop sound delayed correctly until audio thread joins — prevents microphone capture being held open on short recordings (#436)
🔌 IBus / Text Injection
- Engine readiness probe at startup with hardened retries — no more silent IBus failures on first launch (#391)
- Engine instance properly destroyed on keyboard layout switch — fixes silent text injection failure when using Super+Space to switch layouts (reported by @Sbenazar in #388)
- Runtime IBus failure recovery — app now recovers from mid-session IBus crashes without requiring a restart (#411)
📦 Installer / Distro Compatibility
- pywhispercpp library loading repaired — fixes broken installs on systems where the
.sowasn't found at runtime (reported by @mr-bembel in #432, fixed in #433) - openSUSE Tumbleweed dependencies corrected — proper package names and fallback handling (reported and fixed by @peplocanto in #419/#420, additional fallbacks in #418)
- Debian compatibility layer hardened — installer now handles more Debian edge cases robustly (#437)
- Installer validates project config before entering local repo mode — prevents misleading Python version errors (reported by @Jasha10 in #396)
- Existing whisper.cpp builds reused on reinstall — avoids unnecessary multi-minute rebuilds (#421)
- ldconfig refreshed after openSUSE typelib install — fixes runtime library visibility (#438)
- Python 3.14 support +
lxml >= 6.1.0pinned (reported by @mbtw9 in #404)
⚡ whisper.cpp
- CPU thread count reduced — prevents over-subscription on multi-core systems; GPU backend now builds correctly in dev mode (#439)
🔧 Logging
- Runtime log noise cleaned up; hardware detection results are now cached to avoid repeated GPU probes
🔧 Improvements
- Test coverage expanded for recognition internals, IBus edge cases, and CI notification suppression (#410, #414)
- PyPI installation docs clarified with per-distro system dependency requirements (#423)
- Dependencies bumped — Next.js security updates (#399, #429), PostCSS (#2753679) via @dependabot
🙏 Contributors
Huge thanks to everyone who contributed to this release!
Code Contributors
| Contributor | PRs / Commits |
|---|---|
| @jatinkrmalik | #415, #418, #421, #423, #433, #436, #437, #438, #439, #391, #411, #410, #412, #414, + direct commits fixing #388, #396, #401, #404 |
| @peplocanto | #420 — openSUSE dependency fix |
| @juanfradb | #426 — stop sound UX improvement |
| @dependabot | #399, #429 — automated dependency updates |
Issue Reporters & Feature Requestors
Thank you to everyone who took the time to file detailed bug reports and feature requests — this release wouldn't exist without you:
| Contributor | Issue | Summary |
|---|---|---|
| @Sbenazar | #388, #389 | IBus text injection failure on keyboard layout switch |
| @peplocanto | #407, #419 | openSUSE Tumbleweed support and dependency fixes |
| @mr-bembel | #432 | pywhispercpp library loading failure on Ubuntu 26.04 |
| @ericzundel | #363 | Background noise being transcribed |
| @christophtschmidt | #247 | Whisper hallucination of repetitive nonsensical text |
| @bradwood | #393 | Expose whisper.cpp initial_prompt and inference parameters in UI
|
| @sabbari | #401 | Final speech truncated on stop |
| @Jasha10 | #396 | Installer Python version mismatch error |
| @mbtw9 | #404 | lxml >= 6.1.0 compatibility |
📦 Installation
Quick Install (Recommended)
curl -fsSL https://raw.githubusercontent.com/jatinkrmalik/vocalinux/main/install.sh -o /tmp/vl.sh && bash /tmp/vl.shUpdate Existing Installation
Re-run the installer — it will detect your existing setup, preserve your config and models, and update in place:
curl -fsSL https://raw.githubusercontent.com/jatinkrmalik/vocalinux/main/install.sh -o /tmp/vl.sh && bash /tmp/vl.shFrom Source
git clone https://github.com/jatinkrmalik/vocalinux.git
cd vocalinux
git checkout v0.11.0-beta
./install.sh🔍 Full Changelog
Full Diff: v0.10.2-beta...v0.11.0-beta