v0.9.3

Bug Fixes

What's Changed

• fix(models): handle Hugging Face linked size headers by @JavaGT in #1678
• bump tauri version and Mac dock fixes by @cjpais in #1675
• Declutter model cards (only show quantization label on modelcard in debug mode) by @MikaKrul in #1685
• fix: prevent history audio players from playing simultaneously by @j1gc in #1665
• fix(models): don't reload the already-active model when its card is clicked by @kud in #1662

New Contributors

• @JavaGT made their first contribution in #1678
• @j1gc made their first contribution in #1665

Full Changelog: v0.9.2...v0.9.3

v0.9.2

Fix mainly for better GPU detection and selection

What's Changed

• Update descriptions for Moonshine models by @MikaKrul in #1648
• Add cancel download functionality to Onboarding component by @MikaKrul in #1653
• transcribe 0.1.3 by @cjpais in #1664

Full Changelog: v0.9.1...v0.9.2

v0.9.1

Many bug fixes! So many people have contributed. So thank you all for making wonderful contributions to the project.

What's Changed

• ship vsredist directly with the app by @cjpais in #1577
• faster mic initialization by @cjpais in #1582
• fix cyrillic (unicode) path problems by @cjpais in #1187
• bump to transcribe-cpp-0.1.1 by @cjpais in #1589
• Add Dutch (Nederlands) translation by @MikaKrul in #1590
• Update Japanese translations by @niw in #1593
• Improve Dutch (nl) translation accuracy and consistency by @MikaKrul in #1594
• move to auto timestamps for all models by @cjpais in #1602
• fix: gate whisper run extension on model arch, not Feature::InitialPrompt by @btoaldas in #1603
• fix: typo in GigaAM v3 description. by @sergey-v9 in #1613
• Update Italian translations by @albanobattistella in #1604
• fix: show Processing for non-streaming Live overlay by @diligentpig in #1597
• fix: preserve ampersands in custom words by @Piste in #1569
• add openblas to ci and packaging for linux by @cjpais in #1621
• salvage valid settings instead of resetting store on parse failure by @cjpais in #1631
• fix(build): auto-fall-back to AI stub on Command Line Tools-only macOS by @herrerad85 in #1510
• Add Nepali Translation by @Sahilverse in #1632
• handy keys 0.3.0 by @cjpais in #1623
• fix: throttle mic-level IPC to mitigate WebKitWebProcess memory leak by @i3130002 in #1444
• fix(audio): reset resampler state between recordings to prevent audio crosstalk by @andrewleech in #1344
• fix: prevent abort on quit by handling poisoned mutexes in Drop impls by @marcel-dias in #1354
• fix: add prompt injection defense to default post-processing prompt (#1261) by @ChristophNoetel in #1310
• Apply paste delay after key press and increase slider range by @sboys3 in #1465
• bump transcribe.cpp version by @cjpais in #1634
• [bugfix] keep track of current tray icon state by @AlexanderYastrebov in #1158
• fix(tray): log tray icon failures instead of panicking by @rsanheim in #1355
• fix: cancel stalled post-processing without locking shortcuts by @diligentpig in #1614
• fix: tray icon invisible on Windows with dark taskbar + light apps (#1423) by @Sahilverse in #1636
• fix: push-to-talk rapidly toggles recording under X11 auto-repeat (defer synthesized releases) by @mvanhorn in #1605
• fix: surface download errors during onboarding by @kirilan in #1522
• feat: add Light / Dark / System appearance selector by @kud in #1599

New Contributors

• @MikaKrul made their first contribution in #1590
• @niw made their first contribution in #1593
• @btoaldas made their first contribution in #1603
• @sergey-v9 made their first contribution in #1613
• @Piste made their first contribution in #1569
• @herrerad85 made their first contribution in #1510
• @Sahilverse made their first contribution in #1632
• @i3130002 made their first contribution in #1444
• @andrewleech made their first contribution in #1344
• @marcel-dias made their first contribution in #1354
• @sboys3 made their first contribution in #1465
• @mvanhorn made their first contribution in #1605
• @kirilan made their first contribution in #1522
• @kud made their first contribution in #1599

Full Changelog: v0.9.0...v0.9.1

v0.9.0

This is a big release! We now have streaming model support! This is due to transcribe.cpp being integrated into the repo. There will be bugs, so please report them and we will do our best to fix them ASAP

A ton of new models are supported now too.

What's Changed

• refactor(nix): remove phases duplicated from cargo-tauri.hook by @xilec in #1335
• docs(agents): make GitHub workflow rules actionable for AI assistants by @xilec in #1393
• refactor(nix): drop redundant LD_LIBRARY_PATH wrapper prefix by @xilec in #1392
• docs: add overlay/pasting issue workaround for Linux to Known Issues by @manato-tajiri in #1426
• docs: complete Portuguese translation for portable updates by @andrefrd in #1422
• docs: add Intel Mac (x86_64) build instructions by @teremo4ek in #1471
• fix(visualizer): scale FFT window to device sample rate by @SLatz18 in #1491
• fix(models): show size for downloaded models by @diligentpig in #1484
• fix: dropdown overflow in post-processing settings by @arogya01 in #1402
• debug live log viewer in app by @cjpais in #1535
• fix: skip post-processing when transcription is empty by @AhmedDlshad007 in #1537
• fix: eliminate WebKitWebProcess memory leak from overlay mic-level events (#1279) by @captivus in #1447
• debug + perf transcribe cli by @cjpais in #1541
• introduce transcribe.cpp by @cjpais in #1529

New Contributors

• @manato-tajiri made their first contribution in #1426
• @andrefrd made their first contribution in #1422
• @teremo4ek made their first contribution in #1471
• @SLatz18 made their first contribution in #1491
• @diligentpig made their first contribution in #1484
• @arogya01 made their first contribution in #1402
• @AhmedDlshad007 made their first contribution in #1537
• @captivus made their first contribution in #1447

Full Changelog: v0.8.3...v0.9.0

v0.8.3

What's Changed

• nix: set ALSA_PLUGIN_DIR via symlinkJoin by @yuxqiu in #1226
• perf: add reasoning_effort passthrough to avoid thinking-mode latency in local models by @luoxi in #1221
• fix(issue 522): surface paste errors as UI toast notification by @odedindi in #1198
• correct typo in library name, from 'transcription-rs' to 'transcribe-rs' by @ChaseCares in #1260
• Fix (linux) overlay problem in kde by @vpsone in #1121
• docs(readme): add Linux startup troubleshooting section by @xilec in #1266
• query gpu async by @cjpais in #1246
• fix(overlay): parse HANDY_NO_GTK_LAYER_SHELL as boolean by @xilec in #1269
• docs: unify CLAUDE.md and AGENTS.md by @xilec in #1272
• feat: add AWS Bedrock (Mantle) as post-processing provider by @jyuros in #1288
• fix: improve German translation quality by @ChristophNoetel in #1292
• refactor(nix): replace manual bindgen env with rustPlatform.bindgenHook by @xilec in #1255
• chore(nix): add bun node_modules normalization scripts by @xilec in #1256
• docs(audio): clarify what the mic-init timing log actually measures by @rsanheim in #1330
• fix(nix): Fix for macOS build for nixpkgs by @xilec in #1316

New Contributors

• @yuxqiu made their first contribution in #1226
• @luoxi made their first contribution in #1221
• @odedindi made their first contribution in #1198
• @ChaseCares made their first contribution in #1260
• @vpsone made their first contribution in #1121
• @jyuros made their first contribution in #1288
• @ChristophNoetel made their first contribution in #1292
• @rsanheim made their first contribution in #1330

Full Changelog: v0.8.2...v0.8.3

v0.8.2

New Model Alert! Handy now supports the new Cohere Transcribe model! Parakeet transcription performance should also be slightly improved and fail less often. Other bug fixes in this release as well

What's Changed

• Add Hebrew language support by @tomerar in #1155
• Use a more conservative OpenCC conversion for zh-Hant transcription by @Roger-Wu in #1171
• docs: add feature freeze notice to CONTRIBUTING.md and fix typo by @collinstevens in #1175
• [bugfix] remove long repeating words by @AlexanderYastrebov in #976
• fix: don't log cloud provider keys by @shaankhosla in #1145
• docs: add manual release signature verification steps by @AICatKing in #1178
• fix(tray): set tray tooltip text on Windows by @tomerar in #1185
• Load the VAD in parallel to ASR model by @jacksongoode in #1177
• upgrade transcribe rs to 0.3.5 by @cjpais in #1188
• Preserve legacy portable installs during updates by @ferologics in #1167
• fix(nix): remove onnxruntime overlay, use nixpkgs native package by @xilec in #1149
• add cohere by @cjpais in #1200
• Update Italian translation by @albanobattistella in #1202
• fix crash on old cpus by @cjpais in #1176

New Contributors

• @tomerar made their first contribution in #1155
• @Roger-Wu made their first contribution in #1171
• @collinstevens made their first contribution in #1175
• @shaankhosla made their first contribution in #1145
• @AICatKing made their first contribution in #1178

Full Changelog: v0.8.1...v0.8.2

v0.8.1

This is mostly a patch build to make sure we are choosing the fastest GPU option

What's Changed

• Add Bulgarian language by @ron7 in #1136
• fix: require magic string in portable marker to prevent false portable mode on scoop installs by @back1ply in #1126
• enumerate and list gpus in the whisper dropdown, improve auto gpu by @cjpais in #1142
• rearrange some ui by @cjpais in #1123

New Contributors

• @ron7 made their first contribution in #1136

Full Changelog: v0.8.0...v0.8.1

v0.8.0

Handy v0.8.0 is out! It should bring speed improvements for the Whisper models (>2x as fast on my Mac!). Slowly the app is becoming more and more stable and as we inch to a 1.0 release there will be a few features starting to take priority. There may be some stability regressions in this release as we upgraded a lot under the hood. Likely the next few minor releases will have bugfixes.

Beyond the Whisper improvements, audio recordings should now be saved even if the transcription fails. This way you can try to transcribe with another model. Warning: This update contains a DB upgrade which is not backwards compatible. If you move back to an older version of Handy after trying this release, it will crash. You will need to delete your DB. If you need your history preserved, I can help you through migrating it back.

In the coming more major releases you should expect:

• more bug fixes
• streaming support
• handy being an API server (so other applications can use handy to transcribe audio)
• being able to transcribe local files
• eventually local post processing with a fine-tuned model

Please be patient as we work to bring these features in!

A hugeeee thanks to all the contributors in this release and especially @VirenMohindra for helping to review PR's and smooth out CI/CD.

What's Changed

• ci: run full build and quality checks on push to main by @VirenMohindra in #1091
• fix(ci): grant contents: write in main-build.yml by @VirenMohindra in #1092
• whisper-rs 0.16.0 and ort.rc12 by @cjpais in #1041
• fix(ci): use sign-binaries: true in main-build.yml by @VirenMohindra in #1094
• docs: add Raycast extension to Integrations section by @mattiacolombomc in #1064
• Update Raycast integration description by @mattiacolombomc in #1100
• improve history performance by @cjpais in #1107
• fix: sha256 verification to prevent corrupt partial download loop by @VirenMohindra in #1095
• Remove outdated CHANGELOG.md by @isc in #1110
• Save recordings before transcription by @ferologics in #1024
• Adding Swedish language translation by @mohammad-albarham in #1113
• fix: show friendly error when no audio input device is available by @MarceloPrado in #1115
• [BUGFIX] Fix model speed/accuracy indicator alignment by @AlexanderYastrebov in #1116

New Contributors

• @mattiacolombomc made their first contribution in #1064
• @isc made their first contribution in #1110
• @mohammad-albarham made their first contribution in #1113
• @MarceloPrado made their first contribution in #1115

Full Changelog: v0.7.12...v0.8.0

v0.7.12

this is a patch build as I broke giga-am in the last release. this fixes that bug. next release will be 0.8.0 :)

What's Changed

• ci: reduce PR check time from ~30 min to ~1 min by @VirenMohindra in #1073
• Update Italian translation by @gdonisi in #1090
• add extra recording buffer by @cjpais in #1089
• upgrade path from old giga-am to new by @cjpais in #1088

Full Changelog: v0.7.11...v0.7.12

v0.7.11

0.7.11 is the last 0.7.x release of Handy. We will be moving to 0.8.0 next. This has a ton of bug fixes. We also introduced Nvidia's Canary model. There are also now experimental toggles for GPU acceleration. Being able to turn on/off Vulkan. We may have also reduced the number of crashes.

Thank you everyone for your contributions. This is an impressive minor release and it couldn't have been done without you. Looking forward to getting Handy 0.8.0 out to everyone!

What's Changed

• Improve Russian translation by @AlexanderYastrebov in #996
• ui: improve scrollbar UI with custom colors and rounded thumb by @arshit09 in #983
• Add model loading failure notifications with i18n support by @cjpais in #997
• fix: overlay not showing on non-primary monitors by @jondeibel in #969
• fix(nix): Fixed nix hash for dependencies, added CI job and script for updating dependencies versions by @xilec in #948
• docs: add note in readme about homebrew install by @avishj in #1011
• fix(nix): replace manual hash management with bun2nix by @xilec in #1021
• default to handy keys on windows by @cjpais in #1029
• unwind instead of crash by @cjpais in #1030
• Add Windows microphone permission onboarding by @ferologics in #991
• Using the whisper initial_prompt field for custom words by @jmcampanini in #1035
• attempt to clean up warnings on build by @cjpais in #1037
• be able to change models from the tray by @cjpais in #1002
• upgrade to handy-keys 0.2.3 by @cjpais in #1039
• Reduce VolumeSlider step size from 10% to 1% by @sscotth in #944
• Add Windows installation with winget to README by @pl4nty in #1046
• fix: keep Windows microphone guidance Windows-only by @ferologics in #1032
• ensure samples don't get dropped by @cjpais in #1043
• Migrate to transcribe-rs-0.3.1 and add Canary support by @intech in #1023
• experimental: pick between cpu/gpu acceleration + enable directml on windows by @cjpais in #1058
• feat(nix): NixOS integration — ALSA_PLUGIN_DIR + nixosModule by @zitongcharliedeng in #1025

New Contributors

• @AlexanderYastrebov made their first contribution in #996
• @arshit09 made their first contribution in #983
• @jondeibel made their first contribution in #969
• @xilec made their first contribution in #948
• @avishj made their first contribution in #1011
• @jmcampanini made their first contribution in #1035
• @sscotth made their first contribution in #944
• @pl4nty made their first contribution in #1046
• @intech made their first contribution in #1023
• @zitongcharliedeng made their first contribution in #1025

Full Changelog: v0.7.10...v0.7.11