Change defaults to text -> piper wav

Remove API key warning. Small piper fixes.
mideind · Dec 16, 2024 · 934a51c · 934a51c
1 parent 96b3f7a
commit 934a51c
Show file tree

Hide file tree

Showing 3 changed files with 13 additions and 12 deletions.
diff --git a/src/icespeak/settings.py b/src/icespeak/settings.py
@@ -94,18 +94,18 @@ class Settings(BaseSettings):
         extra="ignore",
     )
 
-    DEFAULT_VOICE: str = Field(default="Gudrun", description="Default TTS voice if none is requested.")
+    DEFAULT_VOICE: str = Field(default="bui", description="Default TTS voice if none is requested.")
     DEFAULT_VOICE_SPEED: float = Field(
         default=1.0,
         le=MAX_SPEED,
         ge=MIN_SPEED,
         description="Default TTS voice speed.",
     )
     DEFAULT_TEXT_FORMAT: TextFormats = Field(
-        default=TextFormats.SSML,
+        default=TextFormats.TEXT,
         description="Default format to interpret input text as.",
     )
-    DEFAULT_AUDIO_FORMAT: AudioFormats = Field(default=AudioFormats.MP3, description="Default audio output format.")
+    DEFAULT_AUDIO_FORMAT: AudioFormats = Field(default=AudioFormats.WAV, description="Default audio output format.")
 
     AUDIO_DIR: Optional[Path] = Field(
         default=None,

diff --git a/src/icespeak/tts.py b/src/icespeak/tts.py
@@ -82,9 +82,6 @@ def _setup_voices() -> tuple[VoicesT, ServicesT]:
 
 VOICES, SERVICES = _setup_voices()
 
-assert VOICES, """No voices available. Make sure to provide an API key \
-for at least one speech synthesis engine in your ICESPEAK_KEYS_DIR directory."""
-
 
 _T = TypeVar("_T")
 

diff --git a/src/icespeak/voices/piper_tts.py b/src/icespeak/voices/piper_tts.py
@@ -39,13 +39,17 @@
 class PiperTTSVoice(BaseVoice):
     _NAME: str = "Piper"
     _VOICES: ModuleVoicesT = {
-        "bui": {"id": "bui-medium", "lang": "is_IS", "style": "male"},
-        "salka": {"id": "salka-medium", "lang": "is_IS", "style": "female"},
-        "steinn": {"id": "steinn-medium", "lang": "is_IS", "style": "male"},
-        "ugla": {"id": "ugla-medium", "lang": "is_IS", "style": "female"},
+        "bui": {"id": "bui-medium", "lang": "is-IS", "style": "male"},
+        "salka": {"id": "salka-medium", "lang": "is-IS", "style": "female"},
+        "steinn": {"id": "steinn-medium", "lang": "is-IS", "style": "male"},
+        "ugla": {"id": "ugla-medium", "lang": "is-IS", "style": "female"},
     }
     _AUDIO_FORMATS: ModuleAudioFormatsT = frozenset({"pcm", "wav"})
 
+    @override
+    def __init__(self) -> None:
+        self._avail = True
+
     @property
     @override
     def name(self) -> str:
@@ -80,7 +84,7 @@ def text_to_speech(self, text: str, options: TTSOptions, keys_override: Keys | N
             outfile = SETTINGS.get_empty_file(options.audio_format)
             audio_dir = SETTINGS.get_audio_dir()
             voice = self.voices[options.voice]
-            model = f"{voice['lang']}-{voice['id']}"
+            model = f"{voice['lang'].replace('-','_')}-{voice['id']}"
             data_dir = audio_dir / "Piper"
             piper_args = {
                 "model": shlex.quote(str(model)),
@@ -139,4 +143,4 @@ def text_to_speech(self, text: str, options: TTSOptions, keys_override: Keys | N
         return outfile
 
 
-# TODO: Add option to use GPU. This requires onnxruntime-gpu, the --cuda flag and a functioning CUDA environment.
+# NOTE: Possible to add option to use GPU. This requires onnxruntime-gpu, the --cuda flag and a functioning CUDA environment.