Remove experimental whisper feature

Author: TheEnderek0

README.md

@@ -65,7 +65,7 @@ Let's go over it line by line:
 **OPTIONAL** | This is a convenience feature. If specified, this will get inserted at the start of the subtitle, in bold text. In this example you can imagine the line as: "**Elie Simons**: Hey there my name is elie!". If not specified there will be no actor at the start of the line: "Hey there my name is elie!".
 
 #### "cc" "..."
-**OPTIONAL** | The actual subtitle that will be shown when this scene plays. Experimentally, if this is not specified CCAutoVCD will try to recognize what is said in the specified sound file, and transcribe the message. It does so using `OpenAI whisper's "small" model`. It is not recommended though, as often the processing done on the line or other things can confuse the transcriber. It can be useful for when the lines are changing often due to rapid development and such. If OpenAI whisper fails to load this will be replaced with a placeholder.
+**OPTIONAL** | The actual subtitle that will be shown when this scene plays. If not specified, will be replaced with a placeholder.
 
 #### "cc_color" "255,255,255"
 **OPTIONAL** | The color of the whole line. If not specified defaults to white.

requirements_optional.txt

@@ -1 +0,0 @@
-openai-whisper >= 20250625

src/template_class.py

@@ -15,21 +15,6 @@
 
 PLACEHOLDER_TEXT = "CCAutoVCD: No subtitle specified!"
 
-TRANSCRIBER = None
-
-def TryLoadWhisper(model: str):
-    LOGGER.Info(f"Trying to load auto transcriber: OpenAI-Whisper: {model}")
-    try:
-        import whisper
-    except ImportError:
-        LOGGER.Warning("Failed! Auto transcription is not possible!")
-        return False
-    
-    global TRANSCRIBER
-    TRANSCRIBER = whisper.load_model(model)
-    LOGGER.Info("Loaded!")
-    return True
-
 
 class SndscriptTemplate:
     def __init__(self, kv: Keyvalues, ctx):
@@ -52,14 +37,8 @@ def __init__(self, kv: Keyvalues, ctx):
         try:
             self.cc = kv["cc"]
         except:
-            # Auto transcription
-            LOGGER.VInfo(f"Soundscript {self.name} called for auto transcribing.")
-            if not TRANSCRIBER:
-                if not TryLoadWhisper("small"):
-                    self.cc = PLACEHOLDER_TEXT
-
-            
-            self.cc = TRANSCRIBER.transcribe(str(self.ctx.sound / self.wave))["text"]
+            LOGGER.Warning(f"Soundscript {self.name} has no subtitle!")
+            self.cc = PLACEHOLDER_TEXT
 
         try:
             self.map = kv["map"]