| Age | Commit message (Expand) | Author |
|---|
| 5 days | refactor: share the HF cache helpers between qwen and sglomni | historia |
| 5 days | rewrite backend documentation | historia |
| 5 days | readme | historia |
| 6 days | feat: pip install checks for bad crc errors and purges cache automatically | historia |
| 6 days | fix: increase gpu memory for MOSS with gplang-omni to .87 | historia |
| 6 days | feat: gplang-omni reserves 70% of gpu 0 memory, preventing oom errors on cons... | historia |
| 7 days | fix: warn before using a likely too-big chunk size for sglang-omni models | historia |
| 7 days | feat: reserve 20% vram when running higgs with sglang-omni | historia |
| 7 days | feat: add model and voice labels to generate audiobooks status tui | historia |
| 7 days | feat: move clone .wav directory to settings | historia |
| 7 days | fix: remove overlong text in config backend tui | historia |
| 7 days | feat: voice menu combined for tts/clone paths in tui | historia |
| 7 days | feat: sglang-omni model picker on manual server launch | historia |
| 7 days | rename generator to tts-audiobook-generator. remove pointless upgrade flashes. | historia |
| 7 days | fix: ZONOS2 only generating 1024 tokens max | historia |
| 7 days | feat: auto-update via settings | historia |
| 7 days | feat: sglang backend support | historia |
| 8 days | fix: server status for already-running server | historia |
| 8 days | slop refactor/dedup | historia |
| 8 days | fix: spec santizer for glm, outetts, miotts, minimax. | historia |
| 8 days | fix: hide speech-to-speech only models from config wizard | historia |
| 8 days | fix: better errors for generate all models | historia |
| 9 days | feat: generate a book with all installed models to compare | historia |
| 9 days | fix: truncate excess spaces on generate audiobooks tui | historia |
| 9 days | feat: model capabilities lined up in tui menu | historia |
| 10 days | fix: non-clone models correctly supported in tui, restart server when needed | historia |
| 10 days | feat: output log path when audiobook generation fails | historia |
| 10 days | fix: menu highlight numbering in tui | historia |
| 11 days | fix: crash on bad model_specs from audio.cpp, sanitized | historia |
| 11 days | remove: pointless success flashes in tui for downloads | historia |
| 12 days | fix: no .wav directory error formatting | historia |
| 12 days | feat(audio.cpp): if github binary download fails, automatically start buildin... | historia |
| 12 days | fix: better error when no .wav available to clone | historia |
| 12 days | feat: offer to download prebuilt mac/win binaries, new macos build fallback i... | historia |
| 12 days | feat: cli auto-starts and stops locally-managed servers if no --api-url is pa... | historia |
| 12 days | feat: run auto-build audio.cpp scripts for macos and windows | historia |
| 12 days | refactor: overhaul config.py, remove cli default options | historia |
| 12 days | fix: remove backtrace from console output when using incorrect/incomplete cli... | historia |
| 12 days | fix: japanese characters not truncated correctly in tui terminal output | historia |
| 12 days | feat: .wav hints for all voice directory pickers in tui | historia |
| 12 days | fix: system python used instead of venv python, breaking whisper transcription | historia |
| 12 days | remove: unnecessary flash on succesful build | historia |
| 12 days | feat: unified logging with logging_kit.py | historia |
| 12 days | feat: cuda arch detection for shorter builds, build failure detection | historia |
| 12 days | feat: qwen-tts backend widgets updated to same style as audio.cpp | historia |
| 12 days | fix: silently failing whisper transcription | historia |
| 13 days | remove: unnecessary tui text | historia |
| 13 days | fix: can't check checkboxes in tts model page | historia |
| 13 days | feat: nest start/stop server menu in configure backend menu | historia |
| 13 days | feat: flags for converting individual book files --input-book --output-book | historia |
