* feat(settings): run local VocaGateway from Speech Engine (podman-first)
Add an optional Local VocaGateway panel next to Remote Server so Linux users
can start the pinned v0.1.0 compose project with podman (docker fallback),
watch Live/Pairable/Ready status, pair a phone without loopback QR URLs, and
preset remote_api to the gateway transcriptions endpoint. Stop preserves
volumes; Flatpak fails closed; tray offers Stop only when this session manages
the gateway. Default on-device engine stays whisper.cpp.
* fix(gateway-embed): drop nonexistent compose cpu profile
VocaGateway v0.1.0 exposes default service gateway plus native/cuda/vulkan
profiles, not cpu. Start and stop the default service without --profile, keep
the latest image pin rejection, and prefer save_config for the remote_api preset.
* fix(gateway-embed): keep compose --profile cpu per embed contract
v0.1.0 has no cpu profile entry, but Compose still starts the unprofiled
gateway service when --profile cpu is set, and keeps cuda/vulkan/native off.
That matches the frozen desktop embed contract and stays forward-compatible.
* fix(gateway-embed): harden embed path for settings and compose
Keep podman/docker probing off the GTK thread, reject link-local and
loopback public URLs for phone QR, sanitize compose .env image overrides,
and cap bootstrap token reads so a hostile file cannot blow memory.
* fix(gateway-embed): keep LAN toggle and phone QR in sync
When Allow LAN flips while compose is already managed, rewrite publish
host and PUBLIC_URL and recreate the gateway before Pairable comes back.
Also reject default docker/podman bridge addresses for phone QR.
* test(gateway-embed): pin bridge and LAN-gate pairing cases
Cover 172.17.0.2 / link-local not becoming Pairable, and the mid-run
LAN toggle path that must wait for compose republish before advertising
a phone URL.
* style(gateway-embed): black format embed modules
* fix(gateway-embed): reject short-form loopback and localhost.* for QR
Align Pairable/QR host checks with vocagateway desktop-embed policy so
127.1 / localhost.* never become phone-reachable display URLs.
* fix(gateway-embed): detach Settings status listener on destroy
Gateway poll callbacks were left registered after Settings closed, so
idle_add could touch destroyed GTK widgets while compose stayed up.
* fix(gateway-embed): pass host_env to bridge ip probe
_bridge_iface_ips shells out to host ip; without host_env the AST
gate fails and AppImage bundles can hand the wrong GLib to ip.
* fix(packaging): ship gateway_embed in the AppImage
The explicit setuptools packages list omitted the new package, so the
bundle booted from source trees but died with ModuleNotFoundError on
every distro boot test.
* style(packaging): black the gateway_embed package guard
* fix(settings): pack local VocaGateway into the Advanced island
The rebase onto main auto-merged textually onto the simple Speech Model
page. Keep the gateway card on advanced_box, next to Remote Server.
* fix(gateway-embed): treat missing Compose as unavailable
When podman or docker is present but Compose is not, stop advertising Run.
Return empty compose_args, show an install-Compose hint, and cover it with a unit test.
Also add module loggers on the new gateway_embed modules and annotate the
gateway Settings/tray callbacks Greptile flagged. just lock found no lockfile drift.
* fix(gateway-embed): keep Allow LAN bind matched to the switch
Snapshot the LAN flag for start/republish, record what compose actually
got, then republish again if the switch moved mid-flight so 0.0.0.0 never
lingers while Allow LAN reads off.
* fix(gateway-embed): detect leftover compose after Quit
Quit does not stop compose, so a LAN-published gateway can keep running.
Settings and the tray treated that as Stopped because managed_by_us is
only process memory. Probe compose after runtime detection, show Stop
when it is actually up, and let Stop work even if this session did not
start it.
* fix(gateway-embed): restore LAN switch from leftover compose
Quit leaves compose running, so the next launch now reads
VOCAGATEWAY_PUBLISH_HOST from the embed .env, adopts the project as
managed, and keeps Allow LAN honest so republish can close 0.0.0.0.
* fix(gateway-embed): restore .env if LAN republish fails
compose recreate can fail after write_env_file already pointed
VOCAGATEWAY_PUBLISH_HOST at loopback. Restore the previous .env so
orphan adopt still sees a live 0.0.0.0 bind, and snap Allow LAN back
to the running publish mode.
* style(gateway-embed): type helper args and black MagicMock
* refactor(types): move Readable protocol to common_types
AGENTS.md keeps Protocol interfaces in common_types. Pairing now imports
the shared Readable surface instead of a local _Readable helper.
* fix(gateway-embed): satisfy mypy on rebased embed code
* fix(audio): keep duck restore point when a set reports failure
* feat(ui): add floating glowing dictation overlay
Show a bottom-center non-focus overlay while recognition is LISTENING
or PROCESSING so users get a clear visual cue beyond the tray icon.
Optional GtkLayerShell on Wayland; plain always-on-top GTK fallback
elsewhere. Configurable via ui.show_overlay (default on) and Settings.
* fix(test): isolate dictation overlay tests from GTK pollution
Overlay unit tests no longer construct real GTK windows or import
tray_indicator, which was loading gi.repository early and breaking
later tests that mock GI (single-instance and suspend GLib checks).
* docs: add dictation overlay screenshots
Capture the floating glowing orb in listening (green) and processing
(amber) states for the PR and README screenshots section.
* test(ui): raise dictation overlay coverage for Codecov patch gate
Exercise GTK init, draw/animation, positioning, layer-shell fallbacks,
tray overlay wiring, settings toggle, and config edge cases so patch
coverage on the floating overlay clears the 80% Codecov target.
* test(tray): call overlay helpers via unbound methods
Avoid instance-level MagicMock shadowing of set_overlay_enabled under
some import/mock orderings (seen on CI Python 3.10).
* test(tray): inject explicit config mock for overlay helpers
Assign a MagicMock config_manager on the instance so assertions do not
depend on the constructor patch (which can leave a real ConfigManager
under some Python versions / import orders).
* refactor(ui): ponytail pass on dictation overlay
Drop the hand-drawn mic glyph, simplify the Cairo clear/pulse path,
reuse ConfigManager get_bool/set, tighten tray overlay helpers, and
delete string-grep structural tests that proved nothing.
* fix(ui): pin overlay to bottom strip and stop stealing focus
Use a full-width bottom strip (layer-shell left/right/bottom anchors, or
resize+move fallback) so the orb is bottom-center on Wayland. Map the
window once and toggle opacity instead of hide/show so compositors do not
re-activate it and yank focus out of the editor (which broke IBus inject).
* fix(appimage): seed GtkLayerShell typelib for dictation overlay
* fix(ui): address greptile findings on dictation overlay
* fix(appimage): keep layer-shell typelib optional for direct builds
* feat(audio): lower other audio while dictating
Opt-in setting lowers the default PipeWire or PulseAudio sink while the
microphone is open, then restores that exact level. It stays off by default.
* fix(test): keep pytest.approx working when numpy is mocked
Other test modules replace sys.modules["numpy"] with a MagicMock at import
and leave it there. pytest.approx then crashes on isinstance(..., numpy.bool_).
* fix(audio): restore every channel and do not duck if it cannot be undone
A failed restore-point write no longer lowers the sink. PulseAudio puts
each channel back, and a dead microphone restores playback instead of
waiting for an explicit stop. The recognition test patches the real
audio module when an older test left a mock behind.
* fix(test): put the audio-feedback mock back after the ducking test
The recognition test has to patch the real module, but later tests still
expect the mock that test_text_injector installs. Restore that mock when
the test ends, and annotate the new helpers.
* fix(test): annotate the playback-duck tests
AGENTS.md requires a type hint on every new function. These tests were the
ones still missing them.
* fix(audio): skip arming playback duck after recording already failed
* chore: retrigger review after rebase
* feat(text-injection): let config.json pin the injection backend (#476)
Autodetection has to infer whether IBus commits actually reach the
focused application, and it cannot verify that: commit_text() reports
success even when the compositor drops the text. When the inference is
wrong the failure is silent, and dictation appears to work in XWayland
windows while doing nothing in native ones.
VOCALINUX_FORCE_BACKEND already provides an escape hatch, but it is an
environment variable: it covers a single run, is undocumented outside
the test suite, and a user whose compositor is misdetected has to wrap
the launcher in a shell script to make the choice stick.
Add a persistent `text_injection.backend` setting alongside it, accepting
the same ibus/wtype/ydotool/auto values. The environment variable still
wins so a backend can be A/B-tested for one run without touching the
saved configuration, and the startup log now names which of the two
pinned the choice so a stale setting is diagnosable from --debug output.
Reads the file directly rather than through ConfigManager, keeping this
package independent of the UI layer, exactly as _should_copy_to_clipboard
already does. An unreadable or corrupt config falls back to autodetection
instead of blocking startup, and an unrecognised value is ignored with a
warning so a typo cannot silently pin the wrong backend.
Defaults to "auto", so existing installs behave exactly as before.
* fix(text-injection): tell an explicit VOCALINUX_FORCE_BACKEND=auto from unset
The variable is meant to override the saved text_injection.backend pin for a
single run, which is what makes the two paths A/B-testable without editing
config. Setting it to auto -- the value that means "autodetect this time" --
did not do that: the reader collapsed unset and an explicit auto into the same
"auto" return, so the caller read both as "nothing was forced" and fell through
to the saved pin. The pin won, and there was no way to ask for autodetection
for one run short of editing the config file the variable exists to avoid.
Read the variable as three states instead: unset, an explicit auto, or a
backend name. Only the unset state consults config, so an explicit auto now
short-circuits to autodetection.
An unrecognised value is deliberately treated as unset rather than as auto. A
typo means the user failed to override their preference, not that they asked to
discard it, so falling through to a valid saved pin is the kinder reading.
_forced_backend() keeps its signature and its collapsed return, so its existing
callers and tests are untouched; the new three-state reading lives alongside it
and is what the preference resolver consumes.
* feat(text-injection): accept an xdotool pin and warn when ibus overrides the denylist
Two gaps in the pinnable backend set.
xdotool was not accepted, even though on X11 it is the tool that actually does
the typing. Users were told to pin a *Wayland* backend to mean "not IBus", and
get xdotool as a side effect. Naming it directly is what they were reaching for.
On Wayland it needs no special handling: the value skips IBus like any other
non-ibus pin and falls through to the existing XWayland fallback, so no branch
is added to the selection chain.
The accepted values now live in one tuple, because four places listed them --
both readers and both "expected ..." messages -- and a fifth value would have
had to be added to all four. The two messages derive from the tuple, so they
cannot fall behind it.
Pinning ibus also silently bypasses the compositor denylist. That check exists
because a denylisted compositor makes IBus report a commit as delivered while
the text never arrives (#478, #485), so a pin is the one setting that can put a
user back into a failure with no symptom other than nothing happening. The pin
is still honoured -- an override that ignores you is worse than one that warns
-- but it now says so at warning level, names the compositor, and gives the
remedy. It stays quiet when the compositor genuinely bridges IBus, including a
denylisted one running the ibus-wayland bridge, and on X11 where the denylist
does not apply.
The check is evaluated once. The guard chain short-circuits on the pin, so a
pinned run never reaches the existing compositor branch, and this is its first
and only call -- worth knowing because the check can shell out to pgrep or
gdbus. A test asserts the call count so that stays true.
Documents xdotool in the value table and rewrites the X11 paragraph, which
described xdotool as the implicit result of pinning something else rather than
as a value you can name.
* fix(text-injection): say so when the pinned backend is not the one in use
A pin could be quietly ignored. Pinning a tool that is not installed fell
through to autodetection and typed with whatever it found, under a log line
that had already announced the pin was overriding autodetection. Pinning
xdotool on Wayland did nothing at all, since the selection chain reaches
ydotool and wtype first. Pinning ibus where IBus support is unavailable skipped
the whole IBus branch. In each case the user set a preference, saw no
complaint, and got something else.
These are one condition, not three: whatever ended up doing the typing is not
what was pinned. The check compares the pin against the backend read back from
final state, so a cause nobody has thought of yet is covered without another
branch. It runs at the end of construction rather than inside
_check_dependencies, which returns early on two paths and which the wtype probe
can still override afterwards. The reason clause is added only where a generic
check answers it -- the binary is absent, or IBus is unavailable -- and omitted
otherwise rather than guessing.
Reporting the source separately was letting the environment variable and the
config setting disagree. _check_dependencies resolved the backend, then
re-derived which source had won by asking again -- so the environment was
parsed twice per construction, and a malformed value was reported to the user
twice. Resolution now returns the backend and its source together, which fixes
the duplicate warning and means the source named in a log line is the one that
actually chose the backend. _backend_preference() keeps its signature and reads
the first element.
Two log lines also claimed VOCALINUX_FORCE_BACKEND regardless of where the pin
came from, so a config-set backend was reported as an environment variable the
user had never set. They now name the real source.
The new warning is a startup diagnostic, not a live invariant: a tool switched
later by _try_recover_from_fallback is not reported again. Noted in the
docstring so the boundary is deliberate.
* docs(user-guide): file the injection-backend section under Troubleshooting
The section sat under Customization, directly beneath GPU Acceleration --
where the paragraph immediately above it reads "To check which backend is
being used" about Vulkan and CUDA. The word means something else there, and
two adjacent sections using it for different things is a bad place to explain
either.
It also is not a customization option. It exists because a compositor can make
IBus report a commit as delivered while nothing arrives (#478, #485), and
someone hitting that reads Troubleshooting, not Customization. The section
already opens on the symptom, so it needs no rewriting to belong there.
Pure reorder: no wording changes, and the heading text is unchanged so any
existing link still resolves.
* test(text-injection): cover the pin paths that skip the selection chain
_check_dependencies had three uncovered branches, all reachable by pinning a
backend and none of them exercised: the two early returns that hand over to
IBus before a tool is chosen, and the daemonless-ydotool fallback a wtype pin
lands on when wtype is absent. The early returns matter most, since the pin
check runs at the end of construction precisely so those paths cannot skip
it, and nothing proved that.
Adds the ydotool counterpart to the missing-binary case, so a check keyed on
one tool name cannot pass by coincidence; an X11 test for the guide's claim
that any non-ibus pin turns the IBus path off, with a control asserting IBus
is constructed without a pin so the assertion is the pin's doing; and an
end-to-end explicit-auto case, which until now was only proven at the
resolver rather than through selection.
The pin log line said it was "overriding backend autodetection" before
anything had been checked. An unavailable pin falls through to autodetection,
so that line announced an outcome it could not know and then read as
contradicting the "was not applied" warning. It now states the request only.
_run_with_pin gains ydotoold_ready, defaulting to the previous behaviour so
existing callers are unchanged.
* fix(text-injection): report a config file the backend pin cannot be read from
A corrupt config.json disappeared into logger.debug. The setting is
hand-edited -- that is the workflow the user guide documents -- so a stray
comma is a likely way to reach this, and staying quiet left the user with
autodetection and the silent IBus miss the pin was set to avoid, with nothing
to tell that apart from the pin not working. Unknown values already warned;
an unusable file now does too, naming the file and the parse error.
The bare handler was also catching AttributeError by accident. Valid JSON
that is not an object -- a bare list, string, number or null -- reaches the
lookup and raises there rather than at parse time, and so does a
text_injection section that is not an object, where the top level is a
mapping and the section is not. Both are now shape checks rather than
exception handling, which is what lets the handler narrow to OSError and
ValueError without those five shapes escaping it.
Locating the file moves out of the try. It fixes the handler being unable to
name the path it failed on, and it is a deliberate boundary: config_dir()
resolving a path is not a failure this should paper over, so it is no longer
guarded here at all.
A missing config.json stays silent, which is the default install.
* test(text-injection): stop the ibus-pin tests depending on an installed ydotoold
The bridged-compositor and X11 cases asserted that a run logs no warnings at
all. That is broader than what they are about, and _run_ibus_pin patches
shutil.which to claim every tool exists without patching the probe that then
shells out to one, so the real _ensure_ydotoold ran and warned whenever
/usr/bin/ydotoold was absent. The tests passed on a machine that happens to
have it installed and failed on the CI runners that do not.
The assertions now name the warning they are about, and the helper patches
_ensure_ydotoold the way its sibling _run_with_pin already did, so nothing in
this class of test reads the host's installed packages.
* fix(text-injection): report pending IBus initialization state
* docs(text-injection): clarify backend pin behavior
* fix(config): preserve a hand-edited backend across unrelated saves
ConfigManager writes its full in-memory configuration on every save. A user who edits text_injection.backend in config.json while Vocalinux is running therefore loses that choice when an unrelated Settings or tray save writes the stale default back.
Remember only the backend value observed at load or after a successful save. If the in-memory value is unchanged, preserve the current value from disk; if it changed through ConfigManager, treat that as the intentional write. Missing and explicit auto both reset to auto.
The save aborts rather than overwriting a file whose JSON or text_injection shape cannot be read safely. This is deliberately per-key: no recursive configuration merge or injection-layer dependency is added.
* style(text-injection): remove redundant typing suppression
* fix(config): keep invalid UTF-8 config from blocking startup
* fix(config): preserve malformed backend sections on unrelated saves
* fix(config): retain text setting changes after malformed config
* test(text-injection): annotate backend pin test helpers
The transfer ending is not the dialog's work ending. The checksum still has
to run and the engine still has to load the file, seconds of it on a large
model, and until now that window showed a bar parked at 100% with the
downloader's "Complete!" under it and Cancel as the only button. It reads as
a finished download waiting to be confirmed, or as a hang, and the one
button on offer is the opposite of what the user wants to press.
- ModelDownloadDialog: a fraction of 1.0 no longer paints a finished-looking
bar. It bounces the bar, names the step, says the download is done but the
work is not, and greys out the Cancel button, which cannot interrupt
verification or a model load anyway. Only set_complete() says "Complete!"
and offers OK.
- whisper.cpp: report "Verifying model..." before hashing a file that can be
hundreds of megabytes, the same step the Vosk path has always reported, and
"Loading model..." before backend detection and the load. The tray's
download notification reads the same statuses.
- tests: add return annotations on the new finishing-feedback helpers and
tests (Greptile P2).
Co-authored-by: Jatin K Malik <jatinkrmalik@gmail.com>
* fix(site): align Phone chip, Cost copy, and privacy with GA
Use the official Android beta / iOS TestFlight chip on #ecosystem.
Replace Free forever with Free and open source on privacy and open-source.
Disclose vocalinux.com Google Analytics and scope no-telemetry claims to the installed app.
Fixes#867
* fix(site): make privacy GA disclosure match the live gtag
Spell out page views, the client-ID cookie, and IP/device processing for
G-7NBBNJNNQ7. Align Data Collection with PRODUCT.md, drop the hidden
telemetry weasel, and correct config.json plus the voice-storage row.
* refactor(installer): split implementation into sourced modules
* test(installer): harden module path coverage
* fix(installer): qualify interactive privacy and engine claims
The guided installer overstated local-only privacy and ranked
whisper.cpp without evidence. Match PRODUCT.md: on-device local
engines, optional Remote API, and the ~33 language catalog plus
Auto-detect.
---------
Co-authored-by: Jatin K Malik <jatinkrmalik@gmail.com>
* feat(parakeet): add optional verified Orukeet model
* fix(parakeet): recover from invalid release manifests
* chore: mark Orukeet tip verified
Empty signed tip so the ready-bar Verified gate clears on contributor unsigned history.
---------
Co-authored-by: Jatin K Malik <jatinkrmalik@gmail.com>
* feat(language): follow the active keyboard layout while dictating
Adds a "Follow keyboard layout" switch to Settings -> Speech Model ->
Advanced. With it on, the active keyboard layout is re-read at the start
of every dictation and the matching language is used, so someone who
switches layouts to work in two languages gets the right one without
opening Settings.
#796 already seeded the language once on first run, but from the
*configured* primary layout, which never changes as the user switches
between them: a "fr,us" setup reports fr forever. detect_active_keyboard_
layout() reads GNOME's input sources instead. It takes mru-sources rather
than the current index, because that index goes stale -- observed still
reading 0 after a switch to the second layout -- which is why the IBus
injection path already trusts the MRU list (#497, #738).
An active source that is an IBus engine rather than an xkb layout is a
real answer that cannot be mapped, so the language is left to the engine's
own auto-detect. Borrowing the configured xkb layout there would pin
English for someone typing Japanese through mozc-jp.
The mode is a switch over the language picker rather than an entry inside
it, since it is a mode and not a language. While on, the picker is
disabled and shows what the layout resolves to; turning it off pins that.
A language change no longer re-initialises the engine when the loaded
model takes the language per utterance, which is the whole whisper
family, so following a layout costs a settings read rather than a model
reload. VOSK loads one model per language, Parakeet ignores the picker,
and an English-only model has no non-English weights, so Settings refuses
the mode for all three rather than offering a switch that does nothing.
Fixes#821
* fix(language): keep follow mode armed and map IBus engines
Swap English-only whisper.cpp weights to the multilingual sibling while
passing the layout sentinel into reconfigure, so follow mode stays armed.
Map common GNOME IBus engines onto catalogue languages instead of
leaving them unmapped.
* fix(language): stop IBus hyphen false-maps and gate the sibling swap
The last-token hyphen heuristic mapped mozc-us and foo-us to English.
Restrict it to *-jp/*-kr and name Mozc's extra engines as Japanese.
Only whisper.cpp may consult the whisper.cpp download catalog.
* fix(language): keep multilingual weights while following layout
The picker shows the active layout while follow is on, often en-us.
Deriving whisper.cpp .en from that display made Settings refuse the
mode as English-only, then the next auto-apply persisted a concrete
language and disarmed follow. Use LANGUAGE_FOLLOWS_LAYOUT for .en vs
multilingual derivation instead, including on the load path before
the switch is synced.
* fix(language): type follow helpers and log sibling-swap failures
* fix(test): stub follow-layout model rec so mocked psutil does not blow up
follow-mode English-only guard calls get_recommended_whispercpp_model, which
reads RAM through psutil. Full CI leaves a MagicMock in sys.modules["psutil"],
so ram_gb >= 16 raises TypeError. Patch the settings_dialog alias to return a
multilingual tiny rec so the assertion still holds without real RAM.
* fix(language): catch specific errors on layout lookup and sibling swap
* fix(language): swallow more sibling-swap reconfigure failures
ImportError, TypeError, AttributeError, and ChecksumError from the
multilingual swap must not abort dictation or disarm follow mode.
---------
Co-authored-by: Jatin K Malik <jatinkrmalik@gmail.com>
* fix(settings): speech model follow-ups from #834
Keep the Unused downloads expander in sync with search, recommend Faster
Whisper .en weights for English, parse nvidia-smi MiB VRAM, stop simple
mode from clobbering Faster Whisper/Parakeet, and ceil RAM so ~7.4 GiB
counts as 8.
* fix(settings): keep FW .en picker ids and test unused search
Represent Faster Whisper English catalog variants in the Settings size
combo so tray-saved small.en is not rejected and overwritten. Add
behavioral unused-downloads search coverage for match/nonmatch/clear.
* fix(shortcuts): record unmapped F19/F24 and XF86-aliased F13–F23
* fix(shortcuts): keep split-keyboard modifiers while recording
The recorder only looked at the keyboard that sent the function key, so a modifier held on the other half was dropped. A dead device also stayed in the watch list, which left F19/F24 uncapturable until Record was restarted.
Union held keys across every open keyboard before committing, and drop a device on hangup or read error while scanning for a replacement.
* fix(shortcuts): keep capture tests off real input devices
CI opened real /dev/input paths because the harness patched a stale evdev module, so io watches never attached. Patch the live module, and move the capture device protocol to common_types.
* fix(tests): patch the recorder's own GLib globals
test_settings_dialog reimports the dialog module and leaves the replacement in sys.modules. Recorder methods still close over the original dict, so the harness never recorded timeouts or io watches.
* fix(vosk): refuse to load a model that would exceed available memory
vosk.Model() allocates the whole model in one spike during loading,
including the RNNLM rescoring component some models ship, with no way
to check the requirement up front or load incrementally. Walk the
model directory, compare its on-disk size against psutil's available
memory with a 1.5x margin, and raise before calling into the native
loader instead of letting the kernel OOM killer take the process down
mid-load.
Fixes#676
* fix(vosk): match memory safety margin to the documented peak
greptile flagged the guard's own comment: loading can spike to ~2x the
model's on-disk size, but the margin only required 1.5x available memory,
so a load between those two multiples could still trigger the OOM this
guard exists to prevent.
---------
Co-authored-by: Jatin K Malik <jatinkrmalik@gmail.com>
Shared NAT hits the unauthenticated api.github.com quota and the
update check logged a WARNING that looked like a failure. On 403
or 429, look up the latest stable tag from the releases page
JSON, and stay quiet if that also fails. Nightly still refuses
to treat the stable latest tag as a nightly.
Fixes#845
Snap confinement denies /proc/bus/input/devices unless hardware-observe
is connected. Declare the plug, tell users to connect it next to
raw-input, and update the in-app hints.
Fixes#857
* fix(settings): drop Test Dictation scrolled-window stripe
GTK3 ScrolledWindow defaults to an IN etched shadow, which GNOME/Wayland
renders as a thin vertical stripe on the left of the Test Dictation
revealer even when the buffer is empty. Call set_shadow_type(NONE) to
match the other ScrolledWindows in this dialog.
Fixes#847
* Add missing test return annotation
* test(injection): pin the ydotoold probe so local runs match CI
* chore(pytest): filter the three known warnings out of the summary
* fix(pytest): stop hiding save_settings deprecation globally
Keep the ignore on the one test that still exercises the deprecated alias so the rest of the suite can surface new save_settings() calls.
---------
Co-authored-by: Jatin K Malik <jatinkrmalik@gmail.com>
The gsettings fallback for the bogus xkb:us::eng default only fired on
GNOME/Wayland; on GNOME/X11 the default is accepted and written back
into the X keymap, permanently switching the user off their real
layout (Hungarian, Brazilian ABNT2, etc).
_get_gnome_current_source() also had no fallback when mru-sources is
empty (a fresh session with a single input source never populates the
MRU list), so even with the guard fixed the recovery path still
returned None. It now falls back to the static sources list.
Fixes#699
* fix(release): attach the snap to GitHub Releases
v0.17.0 packed the snap, then Store review of uinput failed the job, so
the file never reached the tag. Build and attach first. A Store hold is
a warning, not a failed release. SHA256SUMS covers the .snap.
snap-backfill.yml rebuilds an existing tag, appends the snap to the
manifest, and attests only that file.
* fix(release): keep snap tags out of the shell and fail real Store errors
Pass the tag through the environment like verify-release.yml. Treat only
an allow-installation review as a successful Store publish. snap-backfill
runs verify_release.py after it rewrites SHA256SUMS.
* style(test): format snap-backfill shell-injection guard with black
The family table was going stale on every sibling release. Keep project
links only. Also say honestly that the 0.17.0 snap is still waiting on
Store review of the uinput plug.
Minor on the stable line. Faster Whisper and Parakeet extras, Speech Model simple setup, Snap uinput, Flatpak bundles on the tag, and QA fixes from the release branch.
Fixes tracked in the PR. Post-release follow-ups: #834.
* fix(deps): export and cap every engine extra
* fix(deps): docstring main and tighten export pairing tests
Give check_exports.main the required public docstring, annotate the
dependency export test helpers, and require --extra to sit with its -o
on the same just lock line (whisper pip-compile stays special-cased).
---------
Co-authored-by: Jatin K Malik <jatinkrmalik@gmail.com>
* fix(audio): open 3-8ch HDA mics at native layout (#813)
HDA analog capture often reports 4 channels. Opening below that native
count can abort PortAudio after read with free(): corrupted unsorted
chunks (same class of bug as #666/#673).
Try the reported channel count first when it is between 3 and 8, then
2ch and 1ch. Pulse devices that advertise 32/128 channels still open
stereo-first. Downmix N channels to mono so engines keep seeing 1ch.
Virtual-device filtering for pipewire/default (#629) is unchanged.
* fix(audio): downmix HDA capture from first channel pair
Averaging all N HDA channels halves speech when only the front pair is live.
Keep stereo mean-of-both, and for 3+ channels average channels 0 and 1.
* fix(audio): pick loudest channel for N-ch downmix
First-pair-only downmix silenced mics on ch2+. For 3+ channels, select
the loudest channel by per-buffer mean-square so speech stays at full
level whether it lands on the front or rear pair. Stereo still averages
both channels.
* fix(audio): sticky N-ch downmix channel for open stream
Pin the loudest-channel pick for the life of a capture stream so a
noise burst on another input cannot switch the mic mid-utterance.
* fix(audio): gate sticky downmix lock on speech energy
Don't pin the N-ch sticky channel from a silent/ambient first buffer;
only lock once the loudest channel clears a documented mean-square floor.
* style: isort sticky-lock import in device config tests
* fix(ui): follow OS dark/light on native GTK installs
AppImage already maps portal/gsettings color-scheme into GTK_THEME.
AUR and install.sh use system GTK3, which ignores prefer-dark. Set
gtk-application-prefer-dark-theme from the same signals, and skip when
GTK_THEME is already set so AppImage and user overrides stay in charge.
Fixes#816
* docs: portal no-preference still checks gsettings
The helper already falls back after a portal 0 or a failed Read.
The module doc said that left GTK defaults, which is only true
after gsettings also comes up empty.
The builder action restore-keys is `flatpak-builder-${arch}`. Our key
was `flatpak-builder-${hash}`, which the action stores as
`flatpak-builder-${hash}-x86_64`. That does not start with
`flatpak-builder-x86_64`, so a packaging hash change (a lock refresh,
or merging one) discarded the whole .flatpak-builder cache and
recompiled numpy and pywhispercpp from source, ~18 min instead of ~4.
Put ${{ matrix.arch }} in the key first so the fallback matches. Then
flatpak-builder skips unchanged modules and only rebuilds what the
hash change actually touched.
PRs no longer watch src/**. python-test already covers those, and the
expensive compile does not. push to main still does, so a sandbox
install break fails the default branch.
* fix(text-injection): route XWayland xdotool fallback through clipboard paste
xdotool type simulates keypresses against the active X keyboard layout, so
a character the layout doesn't map comes out as the wrong glyph or garbled
(#657). The native ydotool path already avoids this by pasting via the
clipboard instead of typing (_inject_with_wayland_tool); the XWayland
xdotool fallback (_inject_with_xdotool) never got the same treatment.
Prefer clipboard paste on that fallback too when ydotool is installed,
falling back to raw xdotool type only when no paste tool is available or
the paste itself fails.
Fixes#657
* fix(text-injection): X11 clipboard paste for the XWayland fallback
jatinkrmalik's review on #680 pointed out the AppImage build does not
bundle ydotool, so the prior fix's ydotool clipboard paste never ran on
the actual reported setup; it fell through to the same layout-dependent
xdotool type as before. It also flagged a real paste-target mismatch:
an XWayland window reads the X11 CLIPBOARD selection, not the Wayland
clipboard wl-copy/ydotool write to.
The XWayland fallback now copies via xclip/xsel and pastes with
xdotool key --clearmodifiers ctrl+v when either is installed, waiting
for held modifiers to release first (the same wait already used by the
native ydotool path). ydotool clipboard paste stays as a second
fallback when xclip/xsel are absent; xdotool type is the last resort.
New regression test drives inject_text() end-to-end with no ydotool on
PATH, reproducing the reported AppImage/GNOME log rather than calling
the fallback method directly, so it also covers the recovery check
that could otherwise skip this path.
* fix(text-injection): thread x11_only through clipboard read/clear so restore matches the paste backend
jatinkrmalik's rebase pulled in the layout-aware paste work from #788,
and greptile's review on the merged branch caught a real gap in this
PR's own XWayland path: _copy_to_clipboard already takes x11_only to
write the X11 CLIPBOARD via xclip/xsel (since that is what an xdotool
paste actually reads), but _read_clipboard and _clear_clipboard never
gained the same parameter. On a Wayland host, _read_clipboard prefers
wl-paste, so the "previous clipboard" saved before an XWayland paste
could be read from an unrelated Wayland clipboard, and restoring it
back through _copy_to_clipboard with no x11_only wrote to that same
wrong backend, leaving the X11 CLIPBOARD stuck on the dictated text.
Threads x11_only through both helpers and every restore call site
(the immediate-failure path and the delayed _restore closure), so an
XWayland paste captures, clears, and restores the same X11 CLIPBOARD
selection it wrote to.
* fix(text-injection): annotate clipboard helpers and XWayland paste tests
Greptile P2 on #680: give `_get_clipboard_tools` a return annotation and
mark the new XWayland clipboard-paste regression tests `-> None`.
* fix(text-injection): derive XWayland paste from environment
Drop the ydotool/wl-copy second fallback on WAYLAND_XDOTOOL (wrong
clipboard, and rc=0 is not a paste) and the x11_only flag stack. Paste
picks xclip/xsel + xdotool from environment; user-facing copy stays
on the session clipboard.
* fix(text-injection): always pass host_env() to the paste chord
The AST guard only recognizes a direct host_env() call. The ternary
around it looked like we were handing the AppImage library path to
xdotool.
* test(text-injection): give XWayland injector stubs clipboard restore state
_inject_with_xdotool now pastes first, and that path reads
_clipboard_restore_target. The __new__ helper in test_text_injector_ext
never set it.
* fix(text-injection): skip clipboard restore on a stale paste failure
If a newer paste has already copied its text, the older paste's failure
handler must not write its saved clipboard back. Success restores already
honored the generation counter; the failure path did not.
* fix(text-injection): keep original clipboard across a stale paste failure
Publish the restore target as soon as copy succeeds (first writer wins)
so an overlapping paste inherits it. If the older paste then fails, the
newer restore still writes the pre-first clipboard, not intermediate
dictated text.
---------
Co-authored-by: Jatin K Malik <jatinkrmalik@gmail.com>
* fix(snap): inject into native Wayland apps via ydotool
Edge rev 7 only ships wtype and xdotool. GNOME rejects the virtual
keyboard protocol, so the snap falls back to xdotool and types only
into XWayland windows. Native GTK/Qt apps get nothing.
Stage ydotool, request the uinput plug, and prefer ydotool when wtype
is rejected. Distro ydotool 0.1.x still needs a writable /dev/uinput;
log `snap connect vocalinux:uinput` when the Snap cannot open it.
* docs(snap): v0.16.2 edge has no uinput plug
The store snap (0.16.2, latest/edge rev 7) does not declare uinput.
`sudo snap connect vocalinux:uinput` fails with "has no plug named
uinput" and dictation only reaches XWayland apps. Spell that out in
README, INSTALL, UPDATE, the website changelog, and the store copy so
people do not run the 0.17 connect command against rev 7.
* docs(snap): drop the unconfirmed 0.17 version from uinput notes
Greptile P2: the tree is still 0.16.2 and the release process does not
name the next tag. Tell people to refresh until `snap info` lists the
uinput plug, instead of waiting on a version we have not cut.
core24 desktop-launch rebuilds gdk-pixbuf-loaders.cache from the first
LD_LIBRARY_PATH directory that ends in the arch triplet. Stage-packages
drop a gdk-pixbuf tree without libpixbufloader_svg.so at
$SNAP/usr/lib/<triplet>, ahead of gnome-platform, so About-page SVGs
fail with "Image type svg is not supported".
Fixes the same core24 GTK3 gap documented at:
https://forum.snapcraft.io/t/missing-svg-support-for-gtk3-in-core24/48851
* fix(packaging): derive the Flatpak and AUR dependency sets from the lock
* test(packaging): add return annotations on new packaging tests
Greptile asked for -> None on the new tests, matching AGENTS.md.
* test(packaging): type parametrized Flatpak source arguments
Greptile asked for type hints on the source parameter, matching AGENTS.md.
---------
Co-authored-by: jatinkrmalik <jatinkrmalik@gmail.com>
* feat(shortcuts): allow bare function keys for push-to-talk
Accept F1–F24 shortcuts without a modifier so dictation can use a
dedicated key (e.g. F10) instead of Shift/Ctrl/Alt combos that fire
while typing. Settings shortcut capture records bare function keys too.
* fix(shortcuts): evdev availability for bare function keys
Check EV_KEY capabilities for bare F1-F24 shortcuts instead of an empty
modifier name. Add availability regression tests and type hints on new tests.
* refactor(shortcuts): fold bare F-keys into ShortcutSpec
Teach primary_modifier to return the bare key so parse_shortcut and the
backend stop special-casing empty modifiers. Capture now asks the parser
instead of re-encoding the F-key allow-list, and evdev shares one
capability probe for modifiers and main keys.
---------
Co-authored-by: Jatin K Malik <jatinkrmalik@gmail.com>
Homepage family strip still pinned the old Win tag. Match PRODUCT.md and
the published VocaWin release. Keep the unsigned / SmartScreen line.
Fixes#811.
* fix(tray): refresh PTT icon when AppIndicator reuses a path
Right-Alt push-to-talk turned the tray red only on the first hold.
Later holds dictated fine but left the icon gray until restart.
StatusNotifier hosts often skip a redraw when IconName returns to an
absolute path already used in the session. Nudge IconThemePath before
set_icon_full so gray-red-gray-red actually turns red again.
* test(tray): assert alternating IconThemePath nudge args
Greptile P2: the PTT redraw regression only checked set_icon_theme_path
call count, so identical paths would still pass. Assert the alternating
ICON_DIR / ICON_DIR+os.sep arguments that drive the host reload.
* fix(injection): read WM_CLASS with xprop so xdotool stops dumping core
`xdotool getwindowclassname` hands an uninitialized XClassHint to XFree
when the window carries no WM_CLASS, so it aborts with SIGABRT
("free(): invalid size") and leaves a coredump behind. On KDE Plasma
Wayland that is the normal case, not an edge case: while a native
Wayland client holds focus, KWin points XWayland's _NET_ACTIVE_WINDOW at
a property-less window, and `xdotool getactivewindow` reports it happily.
Every paste-shortcut probe therefore killed a child process and filled
the journal with coredumps. Injection still worked, because the abort
only made the probe return an empty class, so the symptom was silent
except in `coredumpctl`.
read_wm_class() now reads the property with xprop, which handles a
missing WM_CLASS as an ordinary empty result, and falls back to xdotool
only when xprop is absent and the window looks like a real client (it has
a pid or a title). The debug window-info logging in text_injector.py goes
through the same helper.
Reproduced on Fedora 44, Plasma 6 Wayland, xdotool 3.20211022.1:
`xdotool getwindowclassname 0x200000` exits 134 on the window that
`_NET_ACTIVE_WINDOW` points at.
* fix(injection): keep xdotool classname for class-only X11 clients
When xprop is missing, always allow the xdotool classname fallback so a
real terminal that exposes WM_CLASS but no title or _NET_WM_PID still
classifies as a terminal (Ctrl+Shift+V). xprop remains the primary path,
so Plasma Wayland with xprop never calls getwindowclassname.
---------
Co-authored-by: Jatin K Malik <jatinkrmalik@gmail.com>
* docs: polish project documentation for a clearer public surface
Rewrite the README and supporting docs for a more professional first
impression: less emoji and badge noise, install-first structure, and a
real documentation index. Add a Contributor Covenant, tighten SECURITY
and GitHub templates, and make distro/update guides read as product docs
instead of internal status notes.
* docs: split install guide and add changelog plus support pages
Break INSTALL.md into a short primary guide, manual/PyPI package lists,
and troubleshooting. Add root CHANGELOG.md and SUPPORT.md, wire them
into the docs index and README, and drop the outdated screenshot lag note
now that v0.15 assets are on main.
* docs: humanize PR copy and remove em dashes
Strip em/en dashes from the documentation rewritten in this PR, fix
mid-sentence colon fallout from that pass, and smooth a few stiff
phrases so the public docs read more natural.
* docs: list Faster Whisper after rebase onto main
Installer and CLI already accept faster_whisper. Document it next to
the other engines, drop the outdated screenshot lag note, and keep the
on-device wording honest for local vs remote.
* feat(speech_recognition): add EngineType enum and Engine Protocol
Introduce a typed EngineType enum and an Engine Protocol in common_types.py,
plus an engines/ registry so new speech recognition backends can be plugged in
without scattering string comparisons across the manager.
* feat(speech_recognition): add faster-whisper engine backend
Add a faster-whisper engine implementation using CTranslate2/INT8 for fast
CPU inference. Includes model metadata utility, wiring through the recognition
manager, CLI, settings UI, config, optional dependency, and unit tests. Also
updates the suspend-handler test fixture for the new engine attribute.
* style(ui): format faster-whisper imports for isort 8
* test(speech_recognition): expand faster-whisper coverage for codecov
* test(speech_recognition): fix failing faster-whisper coverage tests
- Import missing get_compute_type helper
- Add tests for English-only detection, cache exceptions, and import-error registry path
- Clear _has_torch_cuda lru_cache around exception test to avoid cached True
- Replace broad builtins.__import__ patch with a fake engine module for registry test
- Restore registry module state after test
* test(speech_recognition): add coverage for faster-whisper edge cases
Cover previously missing lines in:
- faster_whisper_model_info.py: HF cache path/none, hf_hub lookup, fallback traversal, torch CUDA true, psutil exception
- faster_whisper_engine.py: model None guard after is_ready()
- engine registry import-error path
* feat(install): add faster_whisper engine option to install.sh
- Add faster_whisper to --engine help text and examples
- Add faster-whisper as interactive option 5 with description
- Install faster-whisper package and verify import at install time
- Add fallback to whisper.cpp if faster-whisper installation fails
- Add smoke test for faster-whisper import
- Update welcome message engine display name
- Note: faster-whisper models download automatically on first use
* refactor(faster_whisper): apply ponytail review cleanups
- Inline FASTER_WHISPER_MODEL_INFO dict literal; drop unused AVAILABLE_MODELS/MODEL_SIZES
- Reuse _has_torch_cuda() in FasterWhisperEngine.device property
- Remove redundant model None guard after is_ready() verification
- Shrink faster-whisper default config in install.sh to engine + model_size
- Update device tests to manage _has_torch_cuda lru_cache
- Remove test for deleted dead-code guard
* fix(faster_whisper): integrate optional engine with current main
Keep whisper_cpp as default and Parakeet as an extra. Put engine selection
in the Advanced island, install faster-whisper via the optional extra, and
refuse unpinned Hugging Face downloads until model_checksums.txt pins them.
* fix(faster_whisper): download verified models and refresh uv.lock
Settings already opened a download dialog for uncached faster-whisper, but
init only raised ChecksumError. Wire a Parakeet-style downloader that stores
bundles under models_dir, verifies each file against model_checksums.txt, and
loads WhisperModel from that path. Regenerate uv.lock for the optional
faster-whisper extra, pin digests for all eleven Systran models, and add the
missing -> None return annotations.
* fix(faster_whisper): clear engine on unload and fix auto-pause fixture
Auto-pause tests stub __init__, so they never set _faster_whisper_engine.
reinitialize_after_resume then AttributeErrors once Test Python runs.
Mirror the suspend-handler fixture, and release the faster-whisper engine
inside unload_model the same way resume already does.
* fix(faster_whisper): restore previous engine after a failed switch
reconfigure() selected the new engine and released the live model before
init. A failed faster-whisper download then left the manager in ERROR on
the new backend while Settings reverted the pickers to the saved engine,
so dictation stayed dead until restart. Reload the previous engine from
disk (no download) before re-raising so the UI and live manager agree.
* fix(speech_recognition): restore all live settings after failed switch
Failed reconfigure already put the previous engine back, but VAD, device,
remote API, and whisper.cpp knobs stayed on the attempted values while
config on disk did not. Snapshot every field reconfigure writes and restore
the lot before reloading the old backend. Also bind the real
_init_selected_engine onto the MagicMock in the force_reinit test.
* fix(speech_recognition): restore every live field after failed reconfigure
A failed engine switch already reloaded the previous backend, but Greptile
could still miss that VAD, audio device, remote API, and whisper.cpp knobs
were part of the snapshot. Keep one explicit attr list for snapshot/restore,
including the derived voice-commands flag, so live state matches persisted
config after a failed Settings apply.
* chore(speech_recognition): sign tip after rollback follow-up
* feat(commands): localize punctuation voice commands by language
Voice command phrases were English-only even when recognition language
was set to Italian, French, etc. Merge locale punctuation/line-break
aliases from the recognition language into CommandProcessor, keep
English phrases available, and document the behavior.
Fixes#640
Co-authored-by: Jatin K Malik <jatinkrmalik@users.noreply.github.com>
* fix(commands): keep line breaks and French apostrophe aliases
Attach-left punctuation now eats only horizontal whitespace, so a
localized "new line" followed by a period or comma still inserts a
newline. French aliases include ASCII, U+2019, and backtick apostrophes,
and longest-match still wins so "point" cannot fire inside
"point d'interrogation".
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Jatin K Malik <jatinkrmalik@users.noreply.github.com>
subprocess.run(..., timeout=3) SIGKILLs the ydotool client. ydotoold
keeps any keys it already applied, so virtual Left Ctrl can stay held
until logout. Send a best-effort V-up/Ctrl-up (or a legacy ctrl tap)
after paste failure.
Fixes#658
* fix(inject): resolve paste chord for non-QWERTY layouts
ydotool was sending physical KEY_V (47) for Ctrl+V. On German Neo that
key types p, so paste opened Print instead. Prefer wtype keysyms when
usable; otherwise resolve the evdev key for Latin v via xkbcommon, and
read KDE kxkbrc so Plasma Wayland Neo users are covered.
Fixes#787
* fix(inject): read Plasma active layout from real session sources
kxkbrc has no CurrentLayout/LayoutIndex, so multi-layout Neo always
fell back to us at index 0 and paste still hit KEY_V. Prefer D-Bus
getLayout or layout_memory.xml for the active layout, keep kxkbrc
for the list only, and prefer KDE over leftover GNOME on Plasma.
* fix(audio): prepend silent preroll so sink-wake does not clip cues
Short catalog tones start in a fresh paplay process while PipeWire is
still waking a suspended sink, so the first tens of milliseconds of the
cue are missing. Prepend 100ms of silence in the same WAV, cache the
result by source path, mtime, and preroll length, and fall back to the
original file if the preroll cannot be built. CI mock players keep the
raw path.
Fixes#800
* fix(audio): cache sink-wake preroll without tempfile.mkstemp
Some tests replace sys.modules["tempfile"] with a MagicMock, so
mkstemp unpacking fails and playback falls back to the raw cue.
Name the cache file from a short sha256 of the source path, mtime,
and preroll length, write a sibling .partial, then os.replace it
into place. Reload real tempfile and wave in the audio feedback
tests so they still pass after those mocks leak.
Fixes#800
* fix(tests): pin real wave for preroll without reloading tempfile
Popping tempfile out of sys.modules in the audio feedback fixture
made preroll tests pass, but later whisper tests then bound
TemporaryDirectory from a MagicMock. Import real tempfile and wave
at module top, pin only wave while audio_feedback is reimported,
and bind TemporaryDirectory from the real stdlib if a sibling test
already replaced tempfile.
Fixes#800
* fix(audio): catch specific preroll errors and type test helpers
The new preroll path swallowed every exception, which hid bugs and
disagreed with AGENTS.md. Catch OSError, ValueError, and wave.Error
on the inner partial cleanup and the outer fallback so a bad WAV or
cache write still falls back to the original cue, and other errors
still propagate. Annotate the two test helpers the same review
flagged.
* fix(audio): fall back when truncated cues raise EOFError
wave.open can raise EOFError on a truncated RIFF before playback.
Catch it with the other preroll failures so start/stop still play the
original file.
Fixes#800
* fix(asr): keep Parakeet recognition alive after decode errors
sherpa-onnx create/decode exceptions were escaping the recognition
thread. Catch them like whisper.cpp, annotate the Parakeet helpers, and
cover create_stream failure.
* fix(asr): narrow Parakeet transcription exception handling
Keep create_stream/decode survival, but stop swallowing unrelated
bugs. Same RuntimeError/OSError/ValueError tuple as Whisper.
