feat: add ASR filtering options

custom_components/tencentcloud_asr/stt.py

@@ -10,7 +10,15 @@ from homeassistant.helpers.entity_platform import AddEntitiesCallback
 from .tencentcloud_api import TencentCloudAsrAPi
 from .tencentcloud_api import ModuleSupportLanguage
 from .tencentcloud_api import DefaultModel
-from . import SecretIdKey, SecretKeyKey, ModelKey
+from . import (
+    ConvertNumModeKey,
+    FilterDirtyKey,
+    FilterModalKey,
+    FilterPuncKey,
+    ModelKey,
+    SecretIdKey,
+    SecretKeyKey,
+)
 
 _LOGGER = logging.getLogger(__name__)
 
@@ -32,6 +40,7 @@ class ASRSTT(stt.SpeechToTextEntity):
         self._attr_unique_id = f"tencentcloud_asr_stt"
         self._attr_name = f"TencentCloud Asr STT"
         self.hass = hass
+        self.config_entry = config_entry
 
     @property
     def supported_languages(self) -> list[str]:
@@ -71,9 +80,25 @@ class ASRSTT(stt.SpeechToTextEntity):
             audio += chunk
 
         data = base64.b64encode(audio).decode("utf8", "ignore").strip()
-        model: str = self.hass.data.get(ModelKey, DefaultModel)
-        _LOGGER.debug(f"process_audio_stream transcribe: {data} {model}")
-        ret, result = await self.hass.async_add_executor_job(partial(self.tencentCloudApi.SentenceRecognition, model, data))
+        config = {**self.config_entry.data, **self.config_entry.options}
+        model: str = config.get(ModelKey, DefaultModel)
+        _LOGGER.debug(
+            "process_audio_stream transcribe: audio_bytes=%s model=%s",
+            len(audio),
+            model,
+        )
+        ret, result = await self.hass.async_add_executor_job(
+            partial(
+                self.tencentCloudApi.SentenceRecognition,
+                model,
+                data,
+                len(audio),
+                config.get(FilterDirtyKey, "0"),
+                config.get(FilterModalKey, False),
+                config.get(FilterPuncKey, False),
+                config.get(ConvertNumModeKey, True),
+            )
+        )
         if not ret:
             return stt.SpeechResult(result, stt.SpeechResultState.ERROR)
         if not result: