Merge branch 'main' of https://huggingface.co/spaces/aadnk/whisper-webui into main

app.py

@@ -273,15 +273,21 @@ class WhisperTranscriber:
         if ('task' in decodeOptions):
             task = decodeOptions.pop('task')
 
-        if (vadOptions.vadInitialPromptMode == VadInitialPromptMode.PREPEND_ALL_SEGMENTS or 
-            vadOptions.vadInitialPromptMode == VadInitialPromptMode.PREPREND_FIRST_SEGMENT):
+        initial_prompt_mode = vadOptions.vadInitialPromptMode
+
+        # Set default initial prompt mode
+        if (initial_prompt_mode is None):
+            initial_prompt_mode = VadInitialPromptMode.PREPREND_FIRST_SEGMENT
+
+        if (initial_prompt_mode == VadInitialPromptMode.PREPEND_ALL_SEGMENTS or 
+            initial_prompt_mode == VadInitialPromptMode.PREPREND_FIRST_SEGMENT):
             # Prepend initial prompt
-            prompt_strategy = PrependPromptStrategy(initial_prompt, vadOptions.vadInitialPromptMode)
+            prompt_strategy = PrependPromptStrategy(initial_prompt, initial_prompt_mode)
         elif (vadOptions.vadInitialPromptMode == VadInitialPromptMode.JSON_PROMPT_MODE):
             # Use a JSON format to specify the prompt for each segment
             prompt_strategy = JsonPromptStrategy(initial_prompt)
         else:
-            raise ValueError("Invalid vadInitialPromptMode: " + vadOptions.vadInitialPromptMode)
+            raise ValueError("Invalid vadInitialPromptMode: " + initial_prompt_mode)
 
         # Callable for processing an audio file
         whisperCallable = model.create_callback(language, task, prompt_strategy=prompt_strategy, **decodeOptions)

src/config.py

@@ -41,8 +41,10 @@ class VadInitialPromptMode(Enum):
             return VadInitialPromptMode.PREPREND_FIRST_SEGMENT
         elif normalized == "json_prompt_mode":
             return VadInitialPromptMode.JSON_PROMPT_MODE
-        else:
+        elif normalized is not None and normalized != "":
             raise ValueError(f"Invalid value for VadInitialPromptMode: {s}")
+        else:
+            return None
 
 class ApplicationConfig:
     def __init__(self, models: List[ModelConfig] = [], input_audio_max_duration: int = 600, 

src/prompts/jsonPromptStrategy.py

@@ -1,4 +1,5 @@
 import json
+from typing import Dict
 from src.prompts.abstractPromptStrategy import AbstractPromptStrategy
 
 
@@ -25,7 +26,7 @@ class JsonPromptStrategy(AbstractPromptStrategy):
                     
         """
         parsed_json = json.loads(initial_json_prompt)
-        self.segment_lookup = dict[str, JsonPromptSegment]()
+        self.segment_lookup: Dict[str, JsonPromptSegment] = dict() 
         
         for prompt_entry in parsed_json:
             segment_index = prompt_entry["segment_index"]