TTS: dwa głosy (PL gosia / EN ryan) wybierane po języku wypowiedzi

This commit is contained in:
pali112 committed 2026-11-17 12:16:33 +01:00
1 parent bf2eb1938a
commit 3f8f8bbe66
3 files changed
+25 -5

No files matched your search

+21 -3
View File
@@ -100,7 +100,9 @@ DEFAULT_CONFIG = {
"whisper_language": "auto",
"record_cmd": ["pw-record", "--format=s16", "--rate=16000", "--channels=1"],
"piper": "/opt/piper-tts/piper",
"piper_model": "/home/pali112/.local/share/piper-voices/en_US-lessac-medium.onnx",
"piper_model": "", # stary klucz (fallback); normalnie uzywa _pl / _en ponizej
"piper_model_pl": "/home/pali112/models/pl_PL-gosia-medium.onnx",
"piper_model_en": "/home/pali112/models/en_US-ryan-high.onnx",
"play_cmd": ["pw-play"],
"tts_enabled": True,
}
@@ -852,9 +854,22 @@ class XCoPilot:
finally:
self._speak_lock.release()
def _detect_lang(self, text):
"""Zwraca 'pl' jesli tekst zawiera polskie znaki, inaczej 'en'."""
if any(ch in text for ch in "ąćęłńóśźżĄĆĘŁŃÓŚŹŻ"):
return "pl"
return "en"
def _speak_locked(self, text):
piper = self.config.get("piper", "/opt/piper-tts/piper")
model = self.config.get("piper_model", "")
# Wybierz glos wg jezyka wypowiedzi: pl -> gosia, en -> ryan.
if self._detect_lang(text) == "pl":
model = self.config.get("piper_model_pl") or self.config.get("piper_model", "")
else:
model = self.config.get("piper_model_en") or self.config.get("piper_model", "")
if not model:
xp.log("[XCoPilot] TTS: brak modelu piper (piper_model_pl/_en).")
return
import tempfile
wav = tempfile.mktemp(suffix=".wav", prefix="xcopilot_tts_")
env = self._clean_env()
@@ -988,7 +1003,10 @@ class XCoPilot:
_, self.config["voice_enabled"] = imgui.checkbox("Glos wlaczony (STT)", self.config["voice_enabled"])
_, self.config["tts_enabled"] = imgui.checkbox("Odczytywanie odpowiedzi (TTS)", self.config["tts_enabled"])
_, self.config["whisper_model"] = imgui.input_text("Model whisper", self.config["whisper_model"], 512)
_, self.config["piper_model"] = imgui.input_text("Model piper", self.config["piper_model"], 512)
_, self.config["piper_model_pl"] = imgui.input_text(
"Glos PL (gosia)", self.config["piper_model_pl"], 512)
_, self.config["piper_model_en"] = imgui.input_text(
"Glos EN (ryan)", self.config["piper_model_en"], 512)
imgui.separator()
imgui.text("System prompt (osobowosc i format JSON):")
+2 -1
View File
@@ -14,7 +14,8 @@
"whisper_language": "auto",
"record_cmd": ["pw-record", "--format=s16", "--rate=16000", "--channels=1"],
"piper": "/opt/piper-tts/piper",
"piper_model": "/home/pali112/.local/share/piper-voices/en_US-lessac-medium.onnx",
"piper_model_pl": "/home/pali112/models/pl_PL-gosia-medium.onnx",
"piper_model_en": "/home/pali112/models/en_US-ryan-high.onnx",
"play_cmd": ["pw-play"],
"tts_enabled": true
}