diff --git a/linux/CHANGELOG.md b/linux/CHANGELOG.md
index ac070e3..a84da1c 100644
--- a/linux/CHANGELOG.md
+++ b/linux/CHANGELOG.md
@@ -9,6 +9,20 @@ The version is defined in [`blitztext/__init__.py`](blitztext/__init__.py).
## [Unreleased]
+## [2.03.20] - 2026-06-10
+
+### Added
+- **Wakeword benchmark results table.** Results are now shown in a TreeView
+ with one row per engine per voice: Engine | Wakeword | Voice | Detected |
+ Total | Recall % | False fires | Time. Rows are colour-coded green/orange/red
+ by recall. An aggregate "ALL (N voices)" row is appended per engine.
+
+### Fixed
+- **Section header icons now vertically centred with the headline text.**
+ The `.bt-section` CSS class was inadvertently applied to the icon widget,
+ giving it a 14 px top margin and pushing it down. The image no longer
+ receives that class; a `set_pixel_size(14)` pin ensures consistent sizing.
+
## [2.03.19] - 2026-06-10
### Added
diff --git a/linux/blitztext/__init__.py b/linux/blitztext/__init__.py
index 7b00dd3..aba6989 100644
--- a/linux/blitztext/__init__.py
+++ b/linux/blitztext/__init__.py
@@ -6,4 +6,4 @@ counterpart to the macOS Blitztext menu bar app: it runs natively on the host
(not in a container) so it can type into any application via xdotool.
"""
-__version__ = "2.03.19"
+__version__ = "2.03.20"
diff --git a/linux/blitztext/gtksettings.py b/linux/blitztext/gtksettings.py
index 26ddd47..125dd78 100644
--- a/linux/blitztext/gtksettings.py
+++ b/linux/blitztext/gtksettings.py
@@ -316,14 +316,13 @@ def _infobox(parent: Gtk.Box, text: str) -> Gtk.Box:
def _section_title(parent: Gtk.Box, text: str, margin_top: int = 16,
icon: str = "") -> None:
"""A small bold uppercase section header, with optional symbolic icon."""
- row = Gtk.Box(spacing=5)
+ row = Gtk.Box(spacing=4)
row.set_margin_top(margin_top); row.set_margin_bottom(3)
- row.set_valign(Gtk.Align.CENTER)
row.get_style_context().add_class("bt-section")
if icon:
- img = Gtk.Image.new_from_icon_name(icon, Gtk.IconSize.SMALL_TOOLBAR)
+ img = Gtk.Image.new_from_icon_name(icon, Gtk.IconSize.MENU)
+ img.set_pixel_size(14)
img.set_valign(Gtk.Align.CENTER)
- img.get_style_context().add_class("bt-section")
row.pack_start(img, False, False, 0)
lbl = Gtk.Label(xalign=0.0)
lbl.set_valign(Gtk.Align.CENTER)
@@ -772,8 +771,9 @@ class SettingsDialog:
font-size: small;
letter-spacing: 1px;
color: mix(@theme_fg_color, @theme_bg_color, 0.45);
- margin-top: 14px;
- margin-bottom: 2px;
+}
+.bt-section image {
+ margin: 0;
}
.bt-card {
border-radius: 8px;
@@ -2424,6 +2424,32 @@ notebook.bt-nb tab:checked label {
wrun = Gtk.Button(label="Run wakeword benchmark"); wrun.connect("clicked", self._run_wakeword_bench)
wrun.set_halign(Gtk.Align.START)
page.pack_start(wrun, False, False, 6)
+
+ # Results table: Engine | Wakeword | Voice | Detected | Total | Recall% | False fires | Time
+ # Col indices: 0 1 2 3 4 5 6 7
+ self.wwb_store = Gtk.ListStore(str, str, str, str, str, str, str, str, str)
+ ww_tree = Gtk.TreeView(model=self.wwb_store)
+ ww_tree.set_grid_lines(Gtk.TreeViewGridLines.HORIZONTAL)
+ for title, i, expand in [
+ ("Engine", 0, False),
+ ("Wakeword", 1, False),
+ ("Voice", 2, True),
+ ("Detected", 3, False),
+ ("Total", 4, False),
+ ("Recall %", 5, False),
+ ("False fires", 6, False),
+ ("Time (s)", 7, False)]:
+ r = Gtk.CellRendererText()
+ r.set_property("ellipsize", Pango.EllipsizeMode.END)
+ col = Gtk.TreeViewColumn(title, r, text=i, foreground=8)
+ col.set_resizable(True); col.set_expand(expand)
+ ww_tree.append_column(col)
+ ww_sw = Gtk.ScrolledWindow()
+ ww_sw.set_policy(Gtk.PolicyType.AUTOMATIC, Gtk.PolicyType.AUTOMATIC)
+ ww_sw.set_min_content_height(120)
+ ww_sw.add(ww_tree)
+ page.pack_start(ww_sw, True, True, 4)
+
self.wwb_summary = Gtk.Label(xalign=0.0); self.wwb_summary.set_line_wrap(True)
self.wwb_summary.set_selectable(True)
page.pack_start(self.wwb_summary, False, False, 4)
@@ -2633,6 +2659,8 @@ notebook.bt-nb tab:checked label {
for e in engines:
e.model = ww_override
+ if hasattr(self, "wwb_store"):
+ self.wwb_store.clear()
self.wwb_summary.set_markup("Synthesizing and streaming… this talks to your TTS "
"and wyoming-openwakeword servers.")
@@ -2666,35 +2694,58 @@ notebook.bt-nb tab:checked label {
return False
def _wwbench_done(self, runs: list) -> bool:
+ self.wwb_store.clear()
if not runs:
self.wwb_summary.set_markup(
'No engines configured — add one in the Input tab.')
return False
- # Check if all engines failed to synthesize
if all(er.result.expected == 0 for er in runs):
errs = {u.error for er in runs for u in er.result.utterances if u.error}
hint = (" " + GLib.markup_escape_text(next(iter(errs)))) if errs else ""
self.wwb_summary.set_markup(
f"No utterances synthesized — check the TTS model/endpoint.{hint}")
return False
- lines = []
+
+ GREEN, ORANGE, RED_C = "#34c759", "#ff9f0a", "#ff3b30"
+ summary_parts = []
for er in runs:
r = er.result
- name_esc = GLib.markup_escape_text(er.name)
+ phrase = wakeword_bench.wakeword_phrase(er.model)
if r.expected == 0:
- lines.append(f"{name_esc}: no samples synthesized")
+ self.wwb_store.append([er.name, phrase, "—", "—", "—", "—", "—", "—", RED_C])
+ summary_parts.append(f"{er.name}: no samples")
continue
- colour = "#34c759" if r.recall >= 0.9 and r.false_fires == 0 else (
- "#ff9f0a" if r.recall >= 0.6 else "#ff3b30")
+ # Per-voice rows
by_voice = r.recall_by_voice()
- voice_bits = " ".join(
- f"{GLib.markup_escape_text(v)} {d}/{t}" for v, (d, t) in sorted(by_voice.items()))
- lines.append(
- f"{name_esc} "
- f"Recall {r.recall * 100:.0f}% "
- f"({r.detected}/{r.expected}) · False fires: {r.false_fires} · {r.seconds:.0f}s"
- + (f"\n per voice: {voice_bits}" if len(by_voice) > 1 else ""))
- self.wwb_summary.set_markup("\n".join(lines))
+ filler_fires = {v: sum(u.detections for u in r.filler
+ if u.ok and u.voice == v)
+ for v in by_voice}
+ for voice, (det, tot) in sorted(by_voice.items()):
+ recall_pct = det / tot * 100 if tot else 0.0
+ col = GREEN if recall_pct >= 90 else (ORANGE if recall_pct >= 60 else RED_C)
+ ff = filler_fires.get(voice, 0)
+ self.wwb_store.append([
+ er.name, phrase, voice,
+ str(det), str(tot),
+ f"{recall_pct:.0f}%",
+ str(ff),
+ f"{r.seconds:.1f}",
+ col])
+ # Engine summary row (bold aggregates via colour only)
+ overall_col = GREEN if r.recall >= 0.9 and r.false_fires == 0 else (
+ ORANGE if r.recall >= 0.6 else RED_C)
+ self.wwb_store.append([
+ er.name, phrase, f"ALL ({len(by_voice)} voices)",
+ str(r.detected), str(r.expected),
+ f"{r.recall * 100:.0f}%",
+ str(r.false_fires),
+ f"{r.seconds:.1f}",
+ overall_col])
+ summary_parts.append(
+ f"{er.name}: {r.recall*100:.0f}% recall, {r.false_fires} false fires")
+
+ self.wwb_summary.set_markup(" · ".join(
+ GLib.markup_escape_text(p) for p in summary_parts))
return False
# ===== Manual ===========================================================