TTS config: Add a button to download/delete voices for the piper backend

src/calibre/gui2/tts/config.py

@@ -1,8 +1,24 @@
 #!/usr/bin/env python
 # License: GPLv3 Copyright: 2024, Kovid Goyal <kovid at kovidgoyal.net>
 
-
-from qt.core import QCheckBox, QFormLayout, QLabel, QLocale, QMediaDevices, QSize, QSlider, Qt, QTreeWidget, QTreeWidgetItem, QVBoxLayout, QWidget, pyqtSignal
+from qt.core import (
+    QCheckBox,
+    QFormLayout,
+    QHBoxLayout,
+    QIcon,
+    QLabel,
+    QLocale,
+    QMediaDevices,
+    QPushButton,
+    QSize,
+    QSlider,
+    Qt,
+    QTreeWidget,
+    QTreeWidgetItem,
+    QVBoxLayout,
+    QWidget,
+    pyqtSignal,
+)
 
 from calibre.gui2.tts.types import (
     AudioDeviceId,
@@ -113,10 +129,13 @@ def val(self, v):
 
 class Voices(QTreeWidget):
 
+    voice_changed = pyqtSignal()
+
     def __init__(self, parent=None):
         super().__init__(parent)
         self.setHeaderHidden(True)
         self.system_default_voice = Voice()
+        self.currentItemChanged.connect(self.voice_changed)
 
     def sizeHint(self) -> QSize:
         return QSize(400, 500)
@@ -154,14 +173,23 @@ def lang(langcode):
 
     @property
     def val(self) -> str:
-        voice = self.currentItem().data(0, Qt.ItemDataRole.UserRole)
+        voice = self.current_voice
         return voice.name if voice else ''
 
+    @property
+    def current_voice(self) -> Voice | None:
+        ci = self.currentItem()
+        if ci is not None:
+            return ci.data(0, Qt.ItemDataRole.UserRole)
+
 
 class EngineSpecificConfig(QWidget):
 
+    voice_changed = pyqtSignal()
+
     def __init__(self, parent):
         super().__init__(parent)
+        self.engine_name = ''
         self.l = l = QFormLayout(self)
         devs = QMediaDevices.audioOutputs()
         dad = QMediaDevices.defaultAudioOutput()
@@ -183,6 +211,7 @@ def __init__(self, parent):
         self.audio_device = ad = QComboBox(self)
         l.addRow(_('Output a&udio to:'), ad)
         self.voices = v = Voices(self)
+        v.voice_changed.connect(self.voice_changed)
         la = QLabel(_('V&oices:'))
         la.setBuddy(v)
         l.addRow(la)
@@ -241,6 +270,7 @@ def set_engine(self, engine_name):
         else:
             self.layout().setRowVisible(self.audio_device, False)
         self.rebuild_voices()
+        return metadata
 
     def rebuild_voices(self):
         try:
@@ -269,6 +299,29 @@ def as_settings(self) -> EngineSpecificSettings:
                     break
         return ans
 
+    def voice_action(self):
+        v = self.voices.current_voice
+        if v is None:
+            return
+        metadata = available_engines()[self.engine_name]
+        if not metadata.has_managed_voices:
+            return
+        tts = create_tts_backend(self.engine_name)
+        if tts.is_voice_downloaded(v):
+            tts.delete_voice(v)
+        else:
+            tts.download_voice(v)
+
+    def current_voice_is_downloaded(self) -> bool:
+        v = self.voices.current_voice
+        if v is None:
+            return False
+        metadata = available_engines()[self.engine_name]
+        if not metadata.has_managed_voices:
+            return False
+        tts = create_tts_backend(self.engine_name)
+        return tts.is_voice_downloaded(v)
+
 
 class ConfigDialog(Dialog):
 
@@ -279,12 +332,35 @@ def setup_ui(self):
         self.l = l = QVBoxLayout(self)
         self.engine_choice = ec = EngineChoice(self)
         self.engine_specific_config = esc = EngineSpecificConfig(self)
-        ec.changed.connect(esc.set_engine)
+        ec.changed.connect(self.set_engine)
+        esc.voice_changed.connect(self.update_voice_button)
         l.addWidget(ec)
         l.addWidget(esc)
-        l.addWidget(self.bb)
+        self.voice_button = b = QPushButton(self)
+        b.clicked.connect(self.voice_action)
+        h = QHBoxLayout()
+        l.addLayout(h)
+        h.addWidget(b), h.addStretch(10), h.addWidget(self.bb)
         self.initial_engine_choice = ec.value
-        esc.set_engine(self.initial_engine_choice)
+        self.set_engine(self.initial_engine_choice)
+
+    def set_engine(self, engine_name: str) -> None:
+        metadata = self.engine_specific_config.set_engine(engine_name)
+        self.voice_button.setVisible(metadata.has_managed_voices)
+        self.update_voice_button()
+
+    def update_voice_button(self):
+        b = self.voice_button
+        if self.engine_specific_config.current_voice_is_downloaded():
+            b.setIcon(QIcon.ic('trash.png'))
+            b.setText(_('Remove downloaded voice'))
+        else:
+            b.setIcon(QIcon.ic('download-metadata.png'))
+            b.setText(_('Download voice'))
+
+    def voice_action(self):
+        self.engine_specific_config.voice_action()
+        self.update_voice_button()
 
     @property
     def engine_changed(self) -> bool:

src/calibre/gui2/tts/piper.py

@@ -7,6 +7,7 @@
 import re
 import sys
 from collections import deque
+from contextlib import suppress
 from dataclasses import dataclass
 from itertools import count
 from time import monotonic
@@ -441,12 +442,32 @@ def _default_voice(self) -> Voice:
         lang = canonicalize_lang(lang) or lang
         return self._voice_for_lang.get(lang) or self._voice_for_lang['eng']
 
-    def _ensure_voice_is_downloaded(self, voice: Voice) -> tuple[str, str]:
+    def _paths_for_voice(self, voice: Voice) -> tuple[str, str]:
         fname = voice.engine_data['model_filename']
         model_path = os.path.join(cache_dir(), 'piper-voices', fname)
         config_path = os.path.join(os.path.dirname(model_path), fname + '.json')
+        return model_path, config_path
+
+    def is_voice_downloaded(self, v: Voice) -> bool:
+        if not v.name:
+            v = self._default_voice
+        for path in self._paths_for_voice(v):
+            if not os.path.exists(path):
+                return False
+        return True
+
+    def delete_voice(self, v: Voice) -> None:
+        if not v.name:
+            v = self._default_voice
+        for path in self._paths_for_voice(v):
+            with suppress(FileNotFoundError):
+                os.remove(path)
+
+    def _download_voice(self, voice: Voice, download_even_if_exists: bool = False) -> tuple[str, str]:
+        model_path, config_path = self._paths_for_voice(voice)
         if os.path.exists(model_path) and os.path.exists(config_path):
-            return model_path, config_path
+            if not download_even_if_exists:
+                return model_path, config_path
         os.makedirs(os.path.dirname(model_path), exist_ok=True)
         from calibre.gui2.tts.download import DownloadResources
         d = DownloadResources(_('Downloading voice for Read aloud'), _('Downloading neural network for the {} voice').format(voice.human_name), {
@@ -457,6 +478,14 @@ def _ensure_voice_is_downloaded(self, voice: Voice) -> tuple[str, str]:
             return model_path, config_path
         return '', ''
 
+    def download_voice(self, v: Voice) -> None:
+        if not v.name:
+            v = self._default_voice
+        self._download_voice(v, download_even_if_exists=True)
+
+    def _ensure_voice_is_downloaded(self, voice: Voice) -> tuple[str, str]:
+        return self._download_voice(voice)
+
     def validate_settings(self, s: EngineSpecificSettings, parent: QWidget | None) -> bool:
         self._load_voice_metadata()
         voice = self._voice_name_map.get(s.voice_name) or self._default_voice

src/calibre/gui2/tts/types.py

@@ -38,6 +38,7 @@ class EngineMetadata(NamedTuple):
     can_change_pitch: bool = True
     can_change_volume: bool = True
     voices_have_quality_metadata: bool = False
+    has_managed_voices: bool = False
 
 
 class Quality(Enum):
@@ -218,7 +219,7 @@ def qt_engine_metadata(name: str, human_name: str, desc: str, allows_choosing_au
         ans['piper'] = EngineMetadata('piper', _('The Piper Neural Speech Engine'), _(
             'The "piper" engine can track the currently spoken sentence on screen. It uses a neural network '
             'for natural sounding voices. The neural network is run locally on your computer, it is fairly resource intensive to run.'
-        ), TrackingCapability.Sentence, can_change_pitch=False, voices_have_quality_metadata=True)
+        ), TrackingCapability.Sentence, can_change_pitch=False, voices_have_quality_metadata=True, has_managed_voices=True)
     if islinux:
         from speechd.paths import SPD_SPAWN_CMD
         cmd = os.getenv("SPEECHD_CMD", SPD_SPAWN_CMD)
@@ -281,6 +282,15 @@ def reload_after_configure(self) -> None:
     def validate_settings(self, s: EngineSpecificSettings, parent: QWidget | None) -> bool:
         return True
 
+    def is_voice_downloaded(self, v: Voice) -> bool:
+        return True
+
+    def delete_voice(self, v: Voice) -> None:
+        pass
+
+    def download_voice(self, v: Voice) -> None:
+        pass
+
 
 engine_instances: dict[str, TTSBackend] = {}