Fix verbose double-pass bug in CLI/autotranscript; improve logging

scraibe/autotranscript.py

@@ -144,7 +144,7 @@ class Scraibe:
                 "In LocalAI mode, audio_file must be a file path (str)."
             )
 
-        verbose = kwargs.get("verbose", self.verbose)
+        verbose = kwargs.pop("verbose", self.verbose)
         logger.info("transcribe called for: %s", audio_file)
 
         try:
@@ -193,7 +193,7 @@ class Scraibe:
                 "In LocalAI mode, audio_file must be a file path (str)."
             )
 
-        verbose = kwargs.get("verbose", self.verbose)
+        verbose = kwargs.pop("verbose", self.verbose)
         logger.info("transcript_and_summarize called for: %s", audio_file)
 
         # 1) Get diarized + transcribed result

scraibe/cli.py

@@ -217,6 +217,11 @@ def cli():
 
     set_threads(arg_dict.pop("num_threads"))
 
+    # Read shared values once
+    verbose = arg_dict.pop("verbose_output")
+    language = arg_dict.pop("language")
+    num_speakers = arg_dict.pop("num_speakers")
+
     # Build kwargs for Scraibe (LocalAI-backed)
     class_kwargs = {
         "api_url": arg_dict.pop("localai_api_url"),
@@ -227,7 +232,7 @@ def cli():
         "whisper_type": arg_dict.pop("whisper_type"),
         "dia_model": arg_dict.pop("diarization_directory"),
         "use_auth_token": arg_dict.pop("hf_token"),
-        "verbose": arg_dict.pop("verbose_output"),
+        "verbose": verbose,
     }
 
     logger.info("LocalAI API URL: %s", class_kwargs["api_url"] or os.getenv("LOCALAI_API_URL", "<not set>"))
@@ -244,9 +249,9 @@ def cli():
                 logger.info("Starting 'transcribe' for: %s", audio)
                 out = model.transcribe(
                     audio,
-                    language=arg_dict.pop("language"),
+                    language=language,
-                    verbose=arg_dict.pop("verbose_output"),
+                    verbose=verbose,
-                    num_speakers=arg_dict.pop("num_speakers"),
+                    num_speakers=num_speakers,
                 )
                 basename = audio.split("/")[-1].split(".")[0]
                 path = os.path.join(out_folder, f"{basename}.{out_format}")
@@ -263,9 +268,9 @@ def cli():
                     summarizer_api_url=arg_dict.pop("summarizer_api_url"),
                     summarizer_api_key=arg_dict.pop("summarizer_api_key"),
                     summarizer_model=arg_dict.pop("summarizer_model"),
-                    language=arg_dict.pop("language"),
+                    language=language,
-                    verbose=arg_dict.pop("verbose_output"),
+                    verbose=verbose,
-                    num_speakers=arg_dict.pop("num_speakers"),
+                    num_speakers=num_speakers,
                 )
 
                 transcript_text = result.get("transcript", "")