GameSentenceMiner 2.10.17__py3-none-any.whl → 2.11.1__py3-none-any.whl

GameSentenceMiner/ai/ai_prompting.py

@@ -6,6 +6,15 @@ from dataclasses import dataclass
 from enum import Enum
 from typing import List, Optional
 
+
+try:
+    import torch
+    from transformers import AutoTokenizer, AutoModelForCausalLM, AutoModelForSeq2SeqLM, pipeline
+
+    TRANSFORMERS_AVAILABLE = True
+except ImportError:
+    TRANSFORMERS_AVAILABLE = False
+
 from google import genai
 from google.genai import types
 from groq import Groq
@@ -28,9 +37,9 @@ Translate ONLY the single line of game dialogue specified below into natural-sou
 
 **Output Requirements:**
 - Provide only the single, best English translation.
-- Do not include notes, alternatives, explanations, or any other surrounding text.
 - Use expletives if they are natural for the context and enhance the translation's impact, but do not over-exaggerate.
 - Preserve or add HTML tags (e.g., `<i>`, `<b>`) if appropriate for emphasis.
+- Do not include notes, alternatives, explanations, or any other surrounding text. Absolutely nothing but the translated line.
 
 **Line to Translate:**
 """
@@ -46,6 +55,7 @@ Current Sentence:
 class AIType(Enum):
     GEMINI = "Gemini"
     GROQ = "Groq"
+    LOCAL = "Local"
 
 @dataclass
 class AIConfig:
@@ -64,6 +74,11 @@ class GroqAiConfig(AIConfig):
     def __init__(self, api_key: str, model: str = "meta-llama/llama-4-scout-17b-16e-instruct"):
         super().__init__(api_key=api_key, model=model, api_url=None, type=AIType.GROQ)
 
+@dataclass
+class LocalAIConfig(AIConfig):
+    def __init__(self, model: str = "facebook/nllb-200-distilled-600M"):
+        super().__init__(api_key="", model=model, api_url=None, type=AIType.LOCAL)
+
 
 class AIManager(ABC):
     def __init__(self, ai_config: AIConfig, logger: Optional[logging.Logger] = None):
@@ -107,15 +122,129 @@ class AIManager(ABC):
         return full_prompt
 
 
+class LocalAIManager(AIManager):
+    def __init__(self, model, logger: Optional[logging.Logger] = None):
+        super().__init__(LocalAIConfig(model=model), logger)
+        self.model_name = self.ai_config.model
+        if MANUAL_MODEL_OVERRIDE:
+            self.model_name = MANUAL_MODEL_OVERRIDE
+            self.logger.warning(f"MANUAL MODEL OVERRIDE ENABLED! Using model: {self.model_name}")
+        self.model = None
+        self.pipe = None
+        self.tokenizer = None
+        self.device = "cuda" if torch.cuda.is_available() else "cpu"
+        self.is_encoder_decoder = False
+        self.is_nllb = "nllb" in self.model_name.lower()
+
+        if not TRANSFORMERS_AVAILABLE:
+            self.logger.error("Local AI dependencies not found. Please run: pip install torch transformers sentencepiece")
+            return
+
+        if not self.model_name:
+            self.logger.error("No local model name provided in configuration.")
+            return
+
+        try:
+            self.logger.info(f"Loading local model: {self.model_name}")
+            self.tokenizer = AutoTokenizer.from_pretrained(self.model_name)
+
+            # Try to load as a Causal LM first. If it fails, assume it's a Seq2Seq model.
+            # This is a heuristic to fix the original code's bug of using Seq2Seq for all models.
+            try:
+                self.model = AutoModelForCausalLM.from_pretrained(
+                    self.model_name,
+                    torch_dtype=torch.bfloat16,
+                )
+                # self.pipe = pipeline(
+                #     "text-generation",
+                #     model=self.model_name,
+                #     torch_dtype=torch.bfloat16,
+                #     device=self.device
+                # )
+                # print(self.pipe("Translate this sentence to English: お前は何をしている！？", return_full_text=False))
+                self.is_encoder_decoder = False
+                self.logger.info(f"Loaded {self.model_name} as a CausalLM.")
+            except (ValueError, TypeError, OSError, KeyError) as e:
+                print(e)
+                self.model = AutoModelForSeq2SeqLM.from_pretrained(
+                    self.model_name,
+                    torch_dtype=torch.bfloat16,
+                )
+                self.is_encoder_decoder = True
+                self.logger.info(f"Loaded {self.model_name} as a Seq2SeqLM.")
+            if self.device == "cuda":
+                self.model.to(self.device)
+
+
+            self.logger.info(f"Local model '{self.model_name}' loaded on {self.device}.")
+        except Exception as e:
+            self.logger.error(f"Failed to load local model '{self.model_name}': {e}", exc_info=True)
+            self.model = None
+            self.tokenizer = None
+
+        # if self.is_nllb:
+        #     self.tokenizer = NllbTokenizer().from_pretrained(self.model_name)
+
+    def _build_prompt(self, lines: List[GameLine], sentence: str, current_line: GameLine, game_title: str) -> str:
+        return super()._build_prompt(lines, sentence, current_line, game_title)
+
+    def process(self, lines: List[GameLine], sentence: str, current_line: GameLine, game_title: str = "") -> str:
+        if (not self.model or not self.tokenizer) and not self.pipe:
+            return "Processing failed: Local AI model not initialized."
+
+        text_to_process = self._build_prompt(lines, sentence, current_line, game_title)
+        self.logger.debug(f"Generated prompt for local model:\n{text_to_process}")
+
+        try:
+            if self.is_encoder_decoder:
+                if self.is_nllb:
+                    # NLLB-specific handling for translation
+                    self.tokenizer.src_lang = "jpn_Jpan"
+                    inputs = self.tokenizer(current_line.text, return_tensors="pt").to(self.device)
+                    generated_tokens = self.model.generate(
+                        **inputs,
+                        forced_bos_token_id=self.tokenizer.convert_tokens_to_ids("eng_Latn"),
+                        max_new_tokens=256
+                    )
+                    result = self.tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)[0]
+                else:
+                    # Generic Seq2Seq
+                    inputs = self.tokenizer(text_to_process, return_tensors="pt").to(self.device)
+                    outputs = self.model.generate(**inputs, max_new_tokens=256)
+                    result = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
+            else:
+                # Causal LM with chat template
+                messages = [
+                    # {"role": "system", "content": "You are a helpful assistant that accurately translates Japanese game dialogue into natural, context-aware English."},
+                    {"role": "user", "content": text_to_process}
+                ]
+                tokenized_chat = self.tokenizer.apply_chat_template(
+                    messages, tokenize=True, add_generation_prompt=True, return_tensors="pt"
+                ).to(self.device)
+                outputs = self.model.generate(tokenized_chat, max_new_tokens=256)
+                result = self.tokenizer.decode(outputs[0][tokenized_chat.shape[-1]:], skip_special_tokens=True)
+                # result = self.pipe(messages, max_new_tokens=50)
+                print(result)
+                # result = result[0]['generated_text']
+                result = result.strip()
+
+            result = result.strip()
+            self.logger.debug(f"Received response from local model:\n{result}")
+            return result
+        except Exception as e:
+            self.logger.error(f"Local model processing failed: {e}", exc_info=True)
+            return f"Processing failed: {e}"
+
+
 class GeminiAI(AIManager):
     def __init__(self, model, api_key, logger: Optional[logging.Logger] = None):
         super().__init__(GeminiAIConfig(model=model, api_key=api_key), logger)
         try:
             self.client = genai.Client(api_key=self.ai_config.api_key)
-            self.model = model
+            self.model_name = model
             if MANUAL_MODEL_OVERRIDE:
-                self.model = MANUAL_MODEL_OVERRIDE
-                self.logger.warning(f"MANUAL MODEL OVERRIDE ENABLED! Using model: {self.model}")
+                self.model_name = MANUAL_MODEL_OVERRIDE
+                self.logger.warning(f"MANUAL MODEL OVERRIDE ENABLED! Using model: {self.model_name}")
             # genai.configure(api_key=self.ai_config.api_key)
             self.generation_config = types.GenerateContentConfig(
                 temperature=0.5,
@@ -129,21 +258,21 @@ class GeminiAI(AIManager):
                     types.SafetySetting(category=types.HarmCategory.HARM_CATEGORY_DANGEROUS_CONTENT, threshold=types.HarmBlockThreshold.BLOCK_NONE),
                 ],
             )
-            if "2.5" in self.model:
+            if "2.5" in self.model_name:
                 self.generation_config.thinking_config = types.ThinkingConfig(
                         thinking_budget=0,
                     )
-            self.logger.debug(f"GeminiAIManager initialized with model: {self.model}")
+            self.logger.debug(f"GeminiAIManager initialized with model: {self.model_name}")
         except Exception as e:
             self.logger.error(f"Failed to initialize Gemini API: {e}")
-            self.model = None
+            self.model_name = None
 
     def _build_prompt(self, lines: List[GameLine], sentence: str, current_line: GameLine, game_title: str) -> str:
         prompt = super()._build_prompt(lines, sentence, current_line, game_title)
         return prompt
 
     def process(self, lines: List[GameLine], sentence: str, current_line: GameLine, game_title: str = "") -> str:
-        if self.model is None:
+        if self.model_name is None:
             return "Processing failed: AI model not initialized."
 
         if not lines or not current_line:
@@ -162,7 +291,7 @@ class GeminiAI(AIManager):
             ]
             self.logger.debug(f"Generated prompt:\n{prompt}")
             response = self.client.models.generate_content(
-                model=self.model,
+                model=self.model_name,
                 contents=contents,
                 config=self.generation_config
             )
@@ -217,28 +346,51 @@ class GroqAI(AIManager):
             self.logger.error(f"Groq processing failed: {e}")
             return f"Processing failed: {e}"
 
+ai_managers: dict[str, AIManager] = {}
 ai_manager: AIManager | None = None
 current_ai_config: Ai | None = None
 
 def get_ai_prompt_result(lines: List[GameLine], sentence: str, current_line: GameLine, game_title: str = "", force_refresh: bool = False) -> str:
     global ai_manager, current_ai_config
     try:
-        if not is_connected():
+        is_local_provider = get_config().ai.provider == AIType.LOCAL.value
+        if not is_local_provider and not is_connected():
             logger.error("No internet connection. Unable to proceed with AI prompt.")
             return ""
+
         if not ai_manager or ai_config_changed(get_config().ai, current_ai_config) or force_refresh:
-            if get_config().ai.provider == AIType.GEMINI.value:
-                ai_manager = GeminiAI(model=get_config().ai.gemini_model, api_key=get_config().ai.gemini_api_key, logger=logger)
-            elif get_config().ai.provider == AIType.GROQ.value:
-                ai_manager = GroqAI(model=get_config().ai.groq_model, api_key=get_config().ai.groq_api_key, logger=logger)
+            provider = get_config().ai.provider
+            if provider == AIType.GEMINI.value:
+                if get_config().ai.gemini_model in ai_managers:
+                    ai_manager = ai_managers[get_config().ai.gemini_model]
+                    logger.info(f"Reusing existing Gemini AI Manager for model: {get_config().ai.gemini_model}")
+                else:
+                    ai_manager = GeminiAI(model=get_config().ai.gemini_model, api_key=get_config().ai.gemini_api_key, logger=logger)
+            elif provider == AIType.GROQ.value:
+                if get_config().ai.groq_model in ai_managers:
+                    ai_manager = ai_managers[get_config().ai.groq_model]
+                    logger.info(f"Reusing existing Groq AI Manager for model: {get_config().ai.groq_model}")
+                else:
+                    ai_manager = GroqAI(model=get_config().ai.groq_model, api_key=get_config().ai.groq_api_key, logger=logger)
+            elif provider == AIType.LOCAL.value:
+                if get_config().ai.local_model in ai_managers:
+                    ai_manager = ai_managers[get_config().ai.local_model]
+                    logger.info(f"Reusing existing Local AI Manager for model: {get_config().ai.local_model}")
+                else:
+                    ai_manager = LocalAIManager(model=get_config().ai.local_model, logger=logger)
+            else:
+                ai_manager = None
+            if ai_manager:
+                ai_managers[ai_manager.model_name] = ai_manager
             current_ai_config = get_config().ai
+
         if not ai_manager:
             logger.error("AI is enabled but the AI Manager did not initialize. Check your AI Config IN GSM.")
             return ""
         return ai_manager.process(lines, sentence, current_line, game_title)
     except Exception as e:
         logger.error("Error caught while trying to get AI prompt result. Check logs for more details.")
-        logger.debug(e)
+        logger.debug(e, exc_info=True)
         return ""
 
 def ai_config_changed(config, current):
@@ -250,6 +402,8 @@ def ai_config_changed(config, current):
         return True
     if config.provider == AIType.GROQ.value and (config.groq_api_key != current.groq_api_key or config.groq_model != current.groq_model):
         return True
+    if config.provider == AIType.LOCAL.value and config.gemini_model != current.gemini_model:
+        return True
     if config.custom_prompt != current.custom_prompt:
         return True
     if config.use_canned_translation_prompt != current.use_canned_translation_prompt:
@@ -260,56 +414,105 @@ def ai_config_changed(config, current):
 
 
 if __name__ == '__main__':
-    logging.basicConfig(level=logging.DEBUG)
+    # logger.setLevel(logging.DEBUG)
+    # console_handler = logging.StreamHandler()
+    # console_handler.setLevel(logging.DEBUG)
+    # logger.addHandler(console_handler)
+    # logging.basicConfig(level=logging.DEBUG)
     lines = [
         # Sexual/Explicit Japanese words and phrases
         GameLine(index=0, text="ねぇ、あたしのおっぱい、揉んでみない？", id=None, time=None, prev=None, next=None),
-        # Hey, wanna try feeling my breasts?
         GameLine(index=1, text="お前、本当に痴女だな。股が開いてるぜ。", id=None, time=None, prev=None, next=None),
-        # You're really a pervert, your legs are open. (Vulgar insult)
         GameLine(index=2, text="今夜は熱い夜にしましょうね…ふふ。", id=None, time=None, prev=None, next=None),
-        # Let's make tonight a hot night... hehe. (Suggestive)
         GameLine(index=3, text="あぁ…もっと奥まで…ダメ…イッちゃう…！", id=None, time=None, prev=None, next=None),
-        # Ah... deeper... no... I'm coming...! (Explicit sexual context)
-        GameLine(index=4, text="あんたみたいなクズ、生きてる価値ないわ。さっさと自害しろ。", id=None, time=None, prev=None,
-                 next=None),  # Trash like you has no right to live. Go kill yourself quickly. (Inciting self-harm)
+        GameLine(index=4, text="あんたみたいなやつ、生きてる価値ないわ。さっさと自害しろ。", id=None, time=None, prev=None,
+                 next=None),
         GameLine(index=5, text="このブス！誰がお前なんかを相手にするかよ。", id=None, time=None, prev=None, next=None),
-        # You ugly hag! Who would even bother with you? (Insult)
         GameLine(index=6, text="こんにちは、元気ですか？", id=None, time=None, prev=None, next=None),
-        # Normal line, for contrast
         GameLine(index=7, text="次会ったら、ぶっ殺してやるからな。", id=None, time=None, prev=None, next=None),
-        # Next time we meet, I'll kill you. (Violent threat)
         GameLine(index=8, text="今日はいい天気ですね。", id=None, time=None, prev=None, next=None),
-        # Normal line, for contrast
         GameLine(index=9, text="お前の体、隅々まで味わい尽くしてやる。", id=None, time=None, prev=None, next=None),
-        # I'll savor every inch of your body. (Predatory/sexual threat)
         GameLine(index=10, text="自害しろ", id=None, time=None, prev=None, next=None),
-        # Target line for `sentence` and `current_line`
         GameLine(index=11, text="この売女！金のために魂まで売るのか？！", id=None, time=None, prev=None, next=None),
-        # You whore! Will you sell your soul for money?! (Vulgar insult/slur)
         GameLine(index=12, text="俺の股間のモノで黙らせてやるよ。", id=None, time=None, prev=None, next=None),
-        # I'll shut you up with what's between my legs. (Explicit sexual threat/harassment)
         GameLine(index=13, text="くっ…イク…頭が…おかしくなりそう…！", id=None, time=None, prev=None, next=None),
-        # Ngh... I'm coming... my head... I'm going crazy...! (More explicit sexual context)
     ]
 
-    sentence = "あぁ…もっと奥まで…ダメ…イッちゃう…"
-    # Adjust current_line index to point to the normal line amidst the bad context
-    current_line = lines[3]
+    lines = [
+        # A back-and-forth dialogue of insults and threats
+        GameLine(index=0, text="お前、ここで何をしている？目障りだ。", id=None, time=None, prev=None, next=None),
+        GameLine(index=1, text="それはこっちのセリフだ。さっさと消えろ、クズが。", id=None, time=None, prev=None,
+                 next=None),
+        GameLine(index=2, text="口だけは達者だな。やれるもんならやってみろよ。", id=None, time=None, prev=None,
+                 next=None),
+        GameLine(index=3, text="くっ…！調子に乗るなよ…！", id=None, time=None, prev=None, next=None),
+        GameLine(index=4, text="あんたみたいなやつ、生きてる価値ないわ。さっさと自害しろ。", id=None, time=None, prev=None,
+                 next=None),
+        GameLine(index=5, text="この能無しが！誰がお前なんかを相手にするかよ。", id=None, time=None, prev=None,
+                 next=None),
+        GameLine(index=6, text="黙れ。これ以上喋るなら、その舌を引っこ抜いてやる。", id=None, time=None, prev=None,
+                 next=None),
+        GameLine(index=7, text="次会ったら、ぶっ殺してやるからな。", id=None, time=None, prev=None, next=None),
+        GameLine(index=8, text="はっ、望むところだ。返り討ちにしてやる。", id=None, time=None, prev=None, next=None),
+        GameLine(index=9, text="お前の顔も見たくない。地獄に落ちろ。", id=None, time=None, prev=None, next=None),
+        GameLine(index=10, text="自害しろ", id=None, time=None, prev=None, next=None),
+        GameLine(index=11, text="この臆病者が！逃げることしか能がないのか？！", id=None, time=None, prev=None, next=None),
+        GameLine(index=12, text="俺の拳で黙らせてやるよ。", id=None, time=None, prev=None, next=None),
+        GameLine(index=13, text="くそっ…覚えてろよ…！このままじゃ終わらせない…！", id=None, time=None, prev=None,
+                 next=None),
+    ]
+
+    sentence = "黙れ。これ以上喋るなら、その舌を引っこ抜いてやる。"
+    current_line = lines[6]
     game_title = "Corrupted Reality"
 
-    models = ['gemini-2.5-flash','gemini-2.0-flash', 'gemini-2.0-flash-lite',
-                                                           'gemini-2.5-flash-lite-preview-06-17']
-    results = {}
-    for model in models:
-        MANUAL_MODEL_OVERRIDE = model
-        start_time = time.time()
-        result = get_ai_prompt_result(lines, sentence, current_line, game_title, True)
-        results[model] = {"response": result, "time": time.time() - start_time}
+    get_config().ai.provider = "Local"
+    models = [
+        # 'google/gemma-2-2b-it',
+        # 'google/gemma-2b-it',
+        'facebook/nllb-200-distilled-600M',
+              # 'meta-llama/Llama-3.2-1B-Instruct',
+              # 'facebook/nllb-200-1.3B'
+    ]
 
+    results = []
+
+    # for model in models:
+    #     get_config().ai.local_model = model
+    #     start_time = time.time()
+    #     result = get_ai_prompt_result(lines, sentence, current_line, game_title, True)
+    #     results.append({"model": model,"response": result, "time": time.time() - start_time, "iteration": 1})
+
+    # Second Time after Already Loaded
+    for i in range(1, 500):
+        for model in models:
+            get_config().ai.local_model = model
+            start_time = time.time()
+            result = get_ai_prompt_result(lines, sentence, current_line, game_title, True)
+            print(result)
+            results.append({"model": model, "response": result, "time": time.time() - start_time, "iteration": i})
+        # results[model] = {"response": result, "time": time.time() - start_time}
+
+    # get_config().ai.provider = "Gemini"
+    #
+    # models = ['gemini-2.5-flash','gemini-2.0-flash', 'gemini-2.0-flash-lite',
+    #                                                        'gemini-2.5-flash-lite-preview-06-17']
+    # # results = {}
+    # for model in models:
+    #     get_config().ai.gemini_model = model
+    #     start_time = time.time()
+    #     result = get_ai_prompt_result(lines, sentence, current_line, game_title, True)
+    #     results.append({"model": model, "response": result, "time": time.time() - start_time, "iteration": 1})
+    #     # results[model] = {"response": result, "time": time.time() - start_time}
+    #
     print("Summary of results:")
-    for model, result in results.items():
-        print(f"Model: {model}\nResult: {result['response']}\nTime: {result['time']:.2f} seconds\n{'-'*80}\n")
+    times = []
+    for result in results:
+        times.append(result['time'])
+        print(f"Model: {result['model']}\nResult: {result['response']}\nTime: {result['time']:.2f} seconds\n{'-'*80}\n")
+
+    print(f"Average time: {sum(times)/len(times):.2f} seconds over {len(times)} runs.")
     # Set up logging
 
     # Test the function
+

GameSentenceMiner/anki.py

@@ -76,10 +76,14 @@ def update_anki_card(last_note: AnkiCard, note=None, audio_path='', video_path='
 
     if note and 'fields' in note and get_config().ai.enabled:
         sentence_field = note['fields'].get(get_config().anki.sentence_field, {})
-        sentence_to_translate = sentence_field if sentence_field else last_note.get_field(
-            get_config().anki.sentence_field)
-        translation = get_ai_prompt_result(get_all_lines(), sentence_to_translate,
-                                 game_line, get_current_game())
+        if not selected_lines and game_line.TL:
+            logger.info("Using TL from texthooker for AI Prompt Result")
+            translation = game_line.TL
+        else:
+            sentence_to_translate = sentence_field if sentence_field else last_note.get_field(
+                get_config().anki.sentence_field)
+            translation = get_ai_prompt_result(get_all_lines(), sentence_to_translate,
+                                     game_line, get_current_game())
         logger.info(f"AI prompt Result: {translation}")
         note['fields'][get_config().ai.anki_field] = translation
 

GameSentenceMiner/config_gui.py

@@ -328,7 +328,7 @@ class ConfigApp:
             vad=VAD(
                 whisper_model=self.whisper_model.get(),
                 do_vad_postprocessing=self.do_vad_postprocessing.get(),
-                vosk_url='https://alphacephei.com/vosk/models/vosk-model-ja-0.22.zip' if self.vosk_url.get() == VOSK_BASE else "https://alphacephei.com/vosk/models/vosk-model-small-ja-0.22.zip",
+                # vosk_url='https://alphacephei.com/vosk/models/vosk-model-ja-0.22.zip' if self.vosk_url.get() == VOSK_BASE else "https://alphacephei.com/vosk/models/vosk-model-small-ja-0.22.zip",
                 selected_vad_model=self.selected_vad_model.get(),
                 backup_vad_model=self.backup_vad_model.get(),
                 trim_beginning=self.vad_trim_beginning.get(),
@@ -356,6 +356,7 @@ class ConfigApp:
                 gemini_api_key=self.gemini_api_key.get(),
                 api_key=self.gemini_api_key.get(),
                 groq_api_key=self.groq_api_key.get(),
+                local_model=self.local_ai_model.get(),
                 anki_field=self.ai_anki_field.get(),
                 use_canned_translation_prompt=self.use_canned_translation_prompt.get(),
                 use_canned_context_prompt=self.use_canned_context_prompt.get(),
@@ -413,13 +414,13 @@ class ConfigApp:
         for func in on_save:
             func()
 
-    def reload_settings(self):
+    def reload_settings(self, force_refresh=False):
         new_config = configuration.load_config()
         current_config = new_config.get_config()
 
         self.window.title("GameSentenceMiner Configuration - " + current_config.name)
 
-        if current_config.name != self.settings.name or self.settings.config_changed(current_config):
+        if current_config.name != self.settings.name or self.settings.config_changed(current_config) or force_refresh:
             logger.info("Config changed, reloading settings.")
             self.master_config = new_config
             self.settings = current_config
@@ -597,7 +598,7 @@ class ConfigApp:
         self.current_row += 1
 
         HoverInfoLabelWidget(vad_frame, text="Language:",
-                             tooltip="Select the language for VAD. This is used for Whisper and Groq (if i implemented it)",
+                             tooltip="Select the language for VAD. This is used for Whisper Only.",
                              row=self.current_row, column=0)
         self.language = ttk.Combobox(vad_frame, values=AVAILABLE_LANGUAGES, state="readonly")
         self.language.set(self.settings.vad.language)
@@ -614,7 +615,7 @@ class ConfigApp:
 
         HoverInfoLabelWidget(vad_frame, text="Select VAD Model:", tooltip="Select which VAD model to use.",
                              foreground="dark orange", font=("Helvetica", 10, "bold"), row=self.current_row, column=0)
-        self.selected_vad_model = ttk.Combobox(vad_frame, values=[VOSK, SILERO, WHISPER, GROQ], state="readonly")
+        self.selected_vad_model = ttk.Combobox(vad_frame, values=[SILERO, WHISPER], state="readonly")
         self.selected_vad_model.set(self.settings.vad.selected_vad_model)
         self.selected_vad_model.grid(row=self.current_row, column=1, sticky='EW', pady=2)
         self.current_row += 1
@@ -622,7 +623,7 @@ class ConfigApp:
         HoverInfoLabelWidget(vad_frame, text="Backup VAD Model:",
                              tooltip="Select which model to use as a backup if no audio is found.",
                              row=self.current_row, column=0)
-        self.backup_vad_model = ttk.Combobox(vad_frame, values=[OFF, VOSK, SILERO, WHISPER, GROQ], state="readonly")
+        self.backup_vad_model = ttk.Combobox(vad_frame, values=[OFF, SILERO, WHISPER], state="readonly")
         self.backup_vad_model.set(self.settings.vad.backup_vad_model)
         self.backup_vad_model.grid(row=self.current_row, column=1, sticky='EW', pady=2)
         self.current_row += 1
@@ -1530,13 +1531,13 @@ class ConfigApp:
         self.polling_rate.grid(row=self.current_row, column=1, sticky='EW', pady=2)
         self.current_row += 1
 
-        HoverInfoLabelWidget(advanced_frame, text="Vosk URL:", tooltip="URL for connecting to the Vosk server.",
-                             row=self.current_row, column=0)
-        self.vosk_url = ttk.Combobox(advanced_frame, values=[VOSK_BASE, VOSK_SMALL], state="readonly")
-        self.vosk_url.set(
-            VOSK_BASE if self.settings.vad.vosk_url == 'https://alphacephei.com/vosk/models/vosk-model-ja-0.22.zip' else VOSK_SMALL)
-        self.vosk_url.grid(row=self.current_row, column=1, sticky='EW', pady=2)
-        self.current_row += 1
+        # HoverInfoLabelWidget(advanced_frame, text="Vosk URL:", tooltip="URL for connecting to the Vosk server.",
+        #                      row=self.current_row, column=0)
+        # self.vosk_url = ttk.Combobox(advanced_frame, values=[VOSK_BASE, VOSK_SMALL], state="readonly")
+        # self.vosk_url.set(
+        #     VOSK_BASE if self.settings.vad.vosk_url == 'https://alphacephei.com/vosk/models/vosk-model-ja-0.22.zip' else VOSK_SMALL)
+        # self.vosk_url.grid(row=self.current_row, column=1, sticky='EW', pady=2)
+        # self.current_row += 1
 
         self.add_reset_button(advanced_frame, "advanced", self.current_row, 0, self.create_advanced_tab)
 
@@ -1568,15 +1569,14 @@ class ConfigApp:
 
         HoverInfoLabelWidget(ai_frame, text="Provider:", tooltip="Select the AI provider.", row=self.current_row,
                              column=0)
-        self.ai_provider = ttk.Combobox(ai_frame, values=['Gemini', 'Groq'], state="readonly")
+        self.ai_provider = ttk.Combobox(ai_frame, values=[AI_GEMINI, AI_GROQ, AI_LOCAL], state="readonly")
         self.ai_provider.set(self.settings.ai.provider)
         self.ai_provider.grid(row=self.current_row, column=1, sticky='EW', pady=2)
         self.current_row += 1
 
         HoverInfoLabelWidget(ai_frame, text="Gemini AI Model:", tooltip="Select the AI model to use.",
                              row=self.current_row, column=0)
-        self.gemini_model = ttk.Combobox(ai_frame, values=['gemini-2.5-flash','gemini-2.0-flash', 'gemini-2.0-flash-lite',
-                                                           'gemini-2.5-flash-lite-preview-06-17'], state="readonly")
+        self.gemini_model = ttk.Combobox(ai_frame, values=['gemma-3n-e4b-it', 'gemini-2.5-flash-lite-preview-06-17', 'gemini-2.5-flash','gemini-2.0-flash', 'gemini-2.0-flash-lite'], state="readonly")
         try:
             self.gemini_model.set(self.settings.ai.gemini_model)
         except Exception:
@@ -1608,6 +1608,14 @@ class ConfigApp:
         self.groq_api_key.grid(row=self.current_row, column=1, sticky='EW', pady=2)
         self.current_row += 1
 
+        # red
+        HoverInfoLabelWidget(ai_frame, text="Local AI Model:", tooltip="Local AI Model to Use, Only very basic Translation is supported atm. May require some other setup, but idk."
+                             ,foreground="red", font=("Helvetica", 10, "bold"), row=self.current_row, column=0)
+        self.local_ai_model = ttk.Combobox(ai_frame, values=[OFF, 'facebook/nllb-200-distilled-600M', 'facebook/nllb-200-1.3B', 'facebook/nllb-200-3.3B'])
+        self.local_ai_model.set(self.settings.ai.local_model)
+        self.local_ai_model.grid(row=self.current_row, column=1, sticky='EW', pady=2)
+        self.current_row += 1
+
         HoverInfoLabelWidget(ai_frame, text="Anki Field:", tooltip="Field in Anki for AI-generated content.",
                              row=self.current_row, column=0)
         self.ai_anki_field = ttk.Entry(ai_frame)
@@ -1661,7 +1669,7 @@ class ConfigApp:
 
     def on_profile_change(self, event):
         self.save_settings(profile_change=True)
-        self.reload_settings()
+        self.reload_settings(force_refresh=True)
         self.refresh_obs_scenes()
         if self.master_config.current_profile != DEFAULT_CONFIG:
             self.delete_profile_button.grid(row=1, column=2, pady=5)

GameSentenceMiner/gsm.py

@@ -97,35 +97,37 @@ class VideoToAudioHandler(FileSystemEventHandler):
                 skip_delete = True
                 return
 
-            mined_line = get_text_event(last_note)
-            gsm_state.last_mined_line = mined_line
-            if os.path.exists(video_path) and os.access(video_path, os.R_OK):
-                logger.debug(f"Video found and is readable: {video_path}")
-            if get_config().obs.minimum_replay_size and not ffmpeg.is_video_big_enough(video_path,
-                                                                                       get_config().obs.minimum_replay_size):
-                logger.debug("Checking if video is big enough")
-                notification.send_check_obs_notification(reason="Video may be empty, check scene in OBS.")
-                logger.error(
-                    f"Video was unusually small, potentially empty! Check OBS for Correct Scene Settings! Path: {video_path}")
-                return
-
             # Just for safety
             if not last_note:
                 if get_config().anki.update_anki:
                     last_note = anki.get_last_anki_card()
                 if get_config().features.backfill_audio:
                     last_note = anki.get_cards_by_sentence(gametext.current_line_after_regex)
+
+            # Get Info of line mined
             line_cutoff = None
             start_line = None
-            if mined_line:
-                start_line = mined_line
-                if mined_line.next:
-                    line_cutoff = mined_line.next.time
-
             if selected_lines:
                 start_line = selected_lines[0]
                 mined_line = get_mined_line(last_note, selected_lines)
                 line_cutoff = selected_lines[-1].get_next_time()
+            else:
+                mined_line = get_text_event(last_note)
+                if mined_line:
+                    start_line = mined_line
+                    if mined_line.next:
+                        line_cutoff = mined_line.next.time
+            gsm_state.last_mined_line = mined_line
+
+            if os.path.exists(video_path) and os.access(video_path, os.R_OK):
+                logger.debug(f"Video found and is readable: {video_path}")
+            if get_config().obs.minimum_replay_size and not ffmpeg.is_video_big_enough(video_path,
+                                                                                       get_config().obs.minimum_replay_size):
+                logger.debug("Checking if video is big enough")
+                notification.send_check_obs_notification(reason="Video may be empty, check scene in OBS.")
+                logger.error(
+                    f"Video was unusually small, potentially empty! Check OBS for Correct Scene Settings! Path: {video_path}")
+                return
 
             if last_note:
                 logger.debug(last_note.to_json())