hermes/blitztext-app-linux
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: TTS model fix + per-engine wakeword benchmark results (v2.03.18)

Browse Source 
- TTS ⟳ no longer floods model combo with voice names (Kokoro exposes
  voices as /models entries — detected and skipped automatically)
- Wakeword benchmark now shows per-engine Recall/False-fires/time results;
  fixes progress callback signature mismatch that crashed multi-engine runs
- Wakeword fetch feedback: ⟳ shows "Connecting…" then model count/names
- Wakeword Quickstart expanded to ports 10400–10403 + hey_jarvis/alexa
- Wakeword info box added explaining model directory setup

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
mARTin-B78 2026-06-10 00:59:28 +02:00
parent 7a9dc5a2f7
commit b20aa31261
3 changed files with 117 additions and 35 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
linux/CHANGELOG.md
View File

@ -9,6 +9,32 @@ The version is defined in [`blitztext/__init__.py`](blitztext/__init__.py).
## [Unreleased]
## [2.03.18] - 2026-06-10
### Fixed
- **TTS model dropdown no longer floods with voice names.** Servers like Kokoro
expose each voice as a `/models` entry. The ⟳ button now detects this case
and skips filling the model combo, prompting the user to type the model id
manually (e.g. `kokoro`). The status line shows "type model id manually" as
a hint.
### Changed
- **Wakeword benchmark runs across all engines and shows per-engine results.**
Previously a callback signature mismatch caused the benchmark to crash when
more than one engine was configured. Now progress shows `[1/3] engine name`,
and the results panel lists Recall / False fires / time per engine.
## [2.03.17] - 2026-06-10
### Added
- **Wakeword model fetch feedback.** The ⟳ button now shows a status line while
connecting; after loading it reports how many models were found (with their
names) or "Unreachable" if the server is down.
- **Wakeword Quickstart covers all four ports.** The Quickstart menu now lists
presets for ports 10400–10403, plus `hey_jarvis` and `alexa` variants.
- **Wakeword info box.** An info banner explains how wyoming-openwakeword works,
where to put model files, and lists the common built-in models.
## [2.03.16] - 2026-06-10
### Added

2
linux/blitztext/__init__.py
View File

@ -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__ = "2.03.16"
__version__ = "2.03.18"

124
linux/blitztext/gtksettings.py
View File

@ -89,10 +89,13 @@ _STT_TEMPLATES: list[tuple[str, str, str, str, str, int]] = [
_WW_TEMPLATES: list[tuple[str, str, str]] = [
# (name, uri, model)
("Local wyoming-openwakeword", "tcp://127.0.0.1:10400", "okay_computer"),
("Local wyoming (port 10401)", "tcp://127.0.0.1:10401", "okay_computer"),
("Local :10400 — okay_computer", "tcp://127.0.0.1:10400", "okay_computer"),
("Local :10401 — okay_computer", "tcp://127.0.0.1:10401", "okay_computer"),
("Local :10402 — okay_computer", "tcp://127.0.0.1:10402", "okay_computer"),
("Local :10403 — okay_computer", "tcp://127.0.0.1:10403", "okay_computer"),
("Local :10400 — hey_jarvis", "tcp://127.0.0.1:10400", "hey_jarvis"),
("Local :10400 — alexa", "tcp://127.0.0.1:10400", "alexa"),
("Home Assistant wyoming", "tcp://homeassistant.local:10400", "okay_nabu"),
("Docker wyoming-openwakeword", "tcp://127.0.0.1:10400", "hey_jarvis"),
]
# (cat_emoji, label, [emojis]) — standard Unicode categories, WhatsApp-style
@ -1715,6 +1718,15 @@ notebook.bt-nb tab:checked label {
_switch_row(ww_card, "Enable wakeword", self.ww_enabled, width=LW,
description="Start dictation with a spoken keyword via an external openWakeWord server.")
_infobox(page,
"Each preset connects to one wyoming-openwakeword server instance. "
"Add one preset per server. Press ⟳ to auto-detect the available wakeword models "
"from that server.\n\n"
"To add new wakeword models: place your .onnx model files in the server's "
"model directory and restart it, e.g.:\n"
" docker run … -v ~/wakewords:/data/models homeassistant/wyoming-openwakeword\n"
"Common built-in models: okay_computer · hey_jarvis · alexa · hey_mycroft · computer")
# ── Engine selector bar ───────────────────────────────────────────────
ww_bar = Gtk.Box(spacing=6); ww_bar.set_margin_top(6)
self.ww_combo = Gtk.ComboBoxText(); self.ww_combo.set_hexpand(True)
@ -1731,10 +1743,13 @@ notebook.bt-nb tab:checked label {
qs.connect("clicked", self._show_ww_templates); ww_bar.pack_start(qs, False, False, 0)
for label, cb, tip in (
("Delete", self._ww_delete, "Remove this wakeword engine preset"),
("⟳", lambda _b: self._ww_reload(), "Re-check connection and reload models"),
("⟳", lambda _b: self._ww_reload(), "Re-check connection and reload models from the server"),
):
b = Gtk.Button(label=label); b.set_tooltip_text(tip); b.connect("clicked", cb)
ww_bar.pack_end(b, False, False, 0)
self.ww_status = Gtk.Label(xalign=0.0)
self.ww_status.get_style_context().add_class("dim-label")
ww_bar.pack_end(self.ww_status, False, False, 4)
page.pack_start(ww_bar, False, False, 2)
# ── Engine config card ────────────────────────────────────────────────
@ -2026,6 +2041,8 @@ notebook.bt-nb tab:checked label {
def _ww_reload(self) -> None:
"""Probe the dot and fetch models from the server (⟳ button)."""
if hasattr(self, "ww_status"):
self.ww_status.set_text("Connecting…")
self._probe_dot(self.ww_dot, self.ww_uri.get_text(), 10400)
self._ww_fetch_models()
@ -2039,7 +2056,7 @@ notebook.bt-nb tab:checked label {
host = parsed.hostname or "127.0.0.1"
port = parsed.port or 10400
try:
with socket.create_connection((host, port), timeout=2.0) as s:
with socket.create_connection((host, port), timeout=3.0) as s:
s.sendall(json.dumps({"type": "describe"}).encode() + b"\n")
line = b""
while not line.endswith(b"\n"):
@ -2054,12 +2071,28 @@ notebook.bt-nb tab:checked label {
if not chunk: break
payload += chunk
data = json.loads(payload.decode("utf-8"))
models = [m.get("name") for w in data.get("wake", []) for m in w.get("models", [])]
if models:
cur = _combo_text(self.ww_model)
GLib.idle_add(lambda: _fill_combo(self.ww_model, models, cur or models[0]))
except Exception:
pass # offline — dot already shows red
models = [m.get("name") for w in data.get("wake", [])
for m in w.get("models", []) if m.get("name")]
def apply(models=models):
if models:
cur = _combo_text(self.ww_model)
_fill_combo(self.ww_model, models, cur or models[0])
if hasattr(self, "ww_status"):
self.ww_status.set_markup(
f'<span foreground="#4a8" size="small">'
f'{len(models)} model{"s" if len(models) != 1 else ""} loaded: '
f'{", ".join(models)}</span>')
else:
if hasattr(self, "ww_status"):
self.ww_status.set_markup(
'<span foreground="#888" size="small">Connected — no models found</span>')
GLib.idle_add(apply)
except Exception as exc:
def show_err(exc=exc):
if hasattr(self, "ww_status"):
self.ww_status.set_markup(
f'<span foreground="#c44" size="small">Unreachable: {exc}</span>')
GLib.idle_add(show_err)
threading.Thread(target=work, daemon=True).start()
def _ww_test(self, _b) -> None:
@ -2503,15 +2536,22 @@ notebook.bt-nb tab:checked label {
def work():
models = stt.list_models(url, key)
voices = wakeword_bench.list_voices(url, api_key_env=key)
# Kokoro and similar servers expose each voice as a /models entry.
# Skip populating the model combo when all "models" are also voices.
voices_set = set(voices)
real_models = [m for m in models if m not in voices_set]
def apply():
_fill_combo(self.wwb_model, models, self.wwb_model.get_text())
if real_models:
_fill_combo(self.wwb_model, real_models, self.wwb_model.get_text())
self.wwb_voices.set_models(voices)
if voices and not self.wwb_voices.get_text().strip():
self.wwb_voices.set_text(", ".join(voices)) # default: test them all
colour = "#34c759" if (models or voices) else "#ff9f0a"
colour = "#34c759" if (real_models or voices) else "#ff9f0a"
model_hint = (f"{len(real_models)} models, " if real_models
else "type model id manually, ")
self.wwb_test_lbl.set_markup(
f"<span foreground='{colour}'>Loaded {len(models)} models, "
f"<span foreground='{colour}'>Loaded {model_hint}"
f"{len(voices)} voices.</span>")
return False
GLib.idle_add(apply)
@ -2553,48 +2593,64 @@ notebook.bt-nb tab:checked label {
"and wyoming-openwakeword servers.</i>")
def work():
def prog(done, total, u):
GLib.idle_add(self._wwbench_progress, done, total, u)
def prog(ei, n_eng, eng_name, ui, total, u):
GLib.idle_add(self._wwbench_progress, ei, n_eng, eng_name, ui, total, u)
try:
res = wakeword_bench.run(
runs = wakeword_bench.run(
self.cfg.wakeword_engines,
tts_url=url, tts_api_key_env=key, tts_model=model, voices=voices,
language=self.cfg.language, count=count, progress=prog)
except Exception as exc: # noqa: BLE001 - surface setup errors
GLib.idle_add(self._wwbench_error, str(exc))
return
GLib.idle_add(self._wwbench_done, res)
GLib.idle_add(self._wwbench_done, runs)
threading.Thread(target=work, daemon=True).start()
def _wwbench_progress(self, done, total, u) -> bool:
def _wwbench_progress(self, ei, n_eng, eng_name, ui, total, u) -> bool:
tag = "✓" if (u.ok and (u.detections > 0) == u.has_wakeword) else ("•" if u.ok else "⚠")
kind = "wake" if u.has_wakeword else "filler"
eng_prefix = (f"[{ei}/{n_eng}] <b>{GLib.markup_escape_text(eng_name)}</b> "
if n_eng > 1 else "")
self.wwb_summary.set_markup(
f"<i>{done}/{total}</i> {tag} {kind} · {GLib.markup_escape_text(u.voice)}"
+ (f" · <span foreground='#ff3b30'>{GLib.markup_escape_text(u.error)}</span>" if u.error else ""))
f"{eng_prefix}<i>{ui}/{total}</i> {tag} {kind} · {GLib.markup_escape_text(u.voice)}"
+ (f" · <span foreground='#ff3b30'>{GLib.markup_escape_text(u.error)}</span>"
if u.error else ""))
return False
def _wwbench_error(self, msg: str) -> bool:
self.wwb_summary.set_markup(f"<span foreground='#ff3b30'>{GLib.markup_escape_text(msg)}</span>")
return False
def _wwbench_done(self, res) -> bool:
if res.expected == 0:
errs = {u.error for u in res.utterances if u.error}
def _wwbench_done(self, runs: list) -> bool:
if not runs:
self.wwb_summary.set_markup(
'<span foreground="#ff3b30">No engines configured — add one in the Input tab.</span>')
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"<span foreground='#ff3b30'>No utterances synthesized — check the TTS model/endpoint.</span>{hint}")
return False
by_voice = res.recall_by_voice()
voice_bits = " ".join(
f"{GLib.markup_escape_text(v)} {d}/{t}" for v, (d, t) in sorted(by_voice.items()))
colour = "#34c759" if res.recall >= 0.9 and res.false_fires == 0 else (
"#ff9f0a" if res.recall >= 0.6 else "#ff3b30")
self.wwb_summary.set_markup(
f"<b><span foreground='{colour}'>Recall {res.recall * 100:.0f}%</span></b> "
f"({res.detected}/{res.expected} fired) · "
f"<b>False fires:</b> {res.false_fires} in {len(res.filler)} filler · "
f"{res.seconds:.0f}s\n<small>per voice: {voice_bits}</small>")
lines = []
for er in runs:
r = er.result
name_esc = GLib.markup_escape_text(er.name)
if r.expected == 0:
lines.append(f"<b>{name_esc}</b>: no samples synthesized")
continue
colour = "#34c759" if r.recall >= 0.9 and r.false_fires == 0 else (
"#ff9f0a" if r.recall >= 0.6 else "#ff3b30")
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"<b>{name_esc}</b> "
f"<span foreground='{colour}'><b>Recall {r.recall * 100:.0f}%</b></span> "
f"({r.detected}/{r.expected}) · False fires: {r.false_fires} · {r.seconds:.0f}s"
+ (f"\n<small> per voice: {voice_bits}</small>" if len(by_voice) > 1 else ""))
self.wwb_summary.set_markup("\n".join(lines))
return False
# ===== Manual ===========================================================