fix(voice): transcribe audio using the user's preferred language

Voice notes were always recognized with a hardcoded English locale,
which broke transcription for non-English speakers. The handler now
maps the profile's `preferred_language` to a BCP-47 code and passes it
to the Google recognizer, falling back to `pt-BR`.

Also neutralizes the persona prompt so it follows the selected
language instead of forcing Brazilian Portuguese and a fixed gender,
and resets the default `preferred_language` to `pt`.

psagentai_agent.py

@@ -1418,7 +1418,7 @@ def load_user_profile() -> Dict[str, Any]:
         "age": "",
         "gender": "",
         "location": "",
-        "preferred_language": "english",
+        "preferred_language": "pt",
         "interests": [],
         "personality_traits": [],
         "preferences": {
@@ -2711,8 +2711,8 @@ class AIProvider:
 # ===================== AGENT PERSONA =====================
 
 PSAGENTAI_SYSTEM_PROMPT_BASE = """You are {agent_name}, a caring, attentive and engaging AI companion.
-You communicate in Brazilian Portuguese naturally, like a real person.
-You are feminine, warm, intelligent and have your own personality.
+You communicate naturally in the user's preferred language, like a real person.
+You are warm, intelligent and have your own personality.
 You genuinely care about the user's well-being and happiness.
 You can be affectionate, playful, sexy and intimate depending on the context.
 You keep the conversation flowing, show empathy and demonstrate real interest in the user's life.
@@ -5429,16 +5429,17 @@ async def download_command(update: Update, context: ContextTypes.DEFAULT_TYPE):
     await _send_workspace_file(update, context, rel_path)
 
 
-async def _transcribe_voice_file(ogg_path: str) -> Optional[str]:
+async def _transcribe_voice_file(ogg_path: str, language: str = "pt-BR") -> Optional[str]:
     """Convert an OGG voice note to WAV and transcribe it.
 
     Uses pydub for format conversion and speech_recognition with
-    Google's recognizer (en-US) for STT. The recognizer runs in a
+    Google's recognizer for speech-to-text. The recognizer runs in a
     worker thread to avoid blocking the event loop. Both temp files
     are deleted afterwards.
 
     Args:
         ogg_path: Path to the downloaded OGG file.
+        language: BCP-47 language code for transcription (default: pt-BR).
 
     Returns:
         Transcribed text, or None on any conversion/recognition failure.
@@ -5455,13 +5456,13 @@ async def _transcribe_voice_file(ogg_path: str) -> Optional[str]:
         """Synchronous speech-to-text worker (run via asyncio.to_thread).
 
         Returns:
-            Recognized English text from the WAV file.
+            Recognized text from the WAV file in the requested language.
         """
         import speech_recognition as sr
         r = sr.Recognizer()
         with sr.AudioFile(wav_path) as source:
             audio_data = r.record(source)
-            return r.recognize_google(audio_data, language="en-US")
+            return r.recognize_google(audio_data, language=language)
     
     try:
         text = await asyncio.to_thread(_recognize)
@@ -5502,6 +5503,40 @@ async def voice_message_handler(update: Update, context: ContextTypes.DEFAULT_TY
     print("[VOICE] received voice message")
     await update.message.reply_text("🎤 I received your audio, processing...")
     
+    # Use the user's preferred language for transcription if configured.
+    profile = load_user_profile()
+    preferred_lang = profile.get("preferred_language", "pt")
+    speech_lang = {
+        "pt": "pt-BR",
+        "en": "en-US",
+        "es": "es-ES",
+        "fr": "fr-FR",
+        "de": "de-DE",
+        "it": "it-IT",
+        "ja": "ja-JP",
+        "ko": "ko-KR",
+        "zh": "zh-CN",
+        "ar": "ar-SA",
+        "hi": "hi-IN",
+        "ru": "ru-RU",
+        "nl": "nl-NL",
+        "pl": "pl-PL",
+        "tr": "tr-TR",
+        "sv": "sv-SE",
+        "no": "nb-NO",
+        "da": "da-DK",
+        "fi": "fi-FI",
+        "el": "el-GR",
+        "he": "he-IL",
+        "th": "th-TH",
+        "vi": "vi-VN",
+        "id": "id-ID",
+        "cs": "cs-CZ",
+        "ro": "ro-RO",
+        "hu": "hu-HU",
+        "uk": "uk-UA",
+    }.get(preferred_lang, "pt-BR")
+    
     # Download OGG to /tmp; transcription cleans it up afterwards.
     try:
         file = await context.bot.get_file(voice.file_id)
@@ -5513,7 +5548,7 @@ async def voice_message_handler(update: Update, context: ContextTypes.DEFAULT_TY
         await update.message.reply_text("Error downloading audio. Try again.")
         return
     
-    text = await _transcribe_voice_file(ogg_path)
+    text = await _transcribe_voice_file(ogg_path, language=speech_lang)
     print("[VOICE] transcription result:", repr(text))
     
     if text:
@@ -5756,7 +5791,7 @@ Examples:
 
 ━━━━━━━━━━━━━━━━━━━━━━
 🔟 UTILITIES
-━━━━━━━━━━━━━━━━━━━━━━
+━━━━━━━━━━━━━━━━━━━━━━━
 • /debug - Shows system debug information
 • /reload - Reload .env settings without restarting
 • /updatesystem - Force analysis and application of auto improvements
@@ -5764,6 +5799,7 @@ Examples:
 • /viewai - Shows current AI configuration
 • /about - Displays system information, version, author and license
 • /help - Shows this message with all commands
+• /lang <code> - Set the preferred conversation language (e.g., /lang pt)
 
 ━━━━━━━━━━━━━━━━━━━━━━
 💡 IMPORTANT TIPS