Upgrade default Whisper model: small/medium → large-v3 for much better Slovenian/Slavic transcription accuracy

app/main.py

@@ -240,7 +240,7 @@ def process_job(job_id):
             # lang: če None ali 'auto', pusti analyze.py auto-detect
             if job.get("lang") and job["lang"] not in ("auto", ""):
                 cmd += ["--lang", job["lang"]]
-            cmd += ["--model", job.get("whisper_model", "small")]
+            cmd += ["--model", job.get("whisper_model", "large-v3")]
 
             proc = subprocess.run(cmd, capture_output=True, text=True)
             srt_from_claude = None  # Pot do SRT iz Claude-popravljenega transcript-a
@@ -334,7 +334,7 @@ def process_job(job_id):
             cmd += ["--lang", chosen_lang]
         if job.get("no_subs"):
             cmd += ["--no-subs"]
-        cmd += ["--model", job.get("whisper_model", "small")]
+        cmd += ["--model", job.get("whisper_model", "large-v3")]
 
         # DEBUG: zapiši natanko kakšen ukaz se izvede
         update_job(job_id, debug_clip_cmd=" ".join(cmd))
@@ -391,7 +391,7 @@ class YouTubeJobIn(BaseModel):
     duration: Optional[float] = 30
     no_subs: bool = False
     subtitle_style: str = "reels"
-    whisper_model: str = "small"
+    whisper_model: str = "large-v3"
     quality: str = "medium"
 
 
@@ -407,7 +407,7 @@ class StartJobIn(BaseModel):
     min_duration: Optional[float] = 20
     no_subs: bool = False
     subtitle_style: str = "reels"
-    whisper_model: str = "small"
+    whisper_model: str = "large-v3"
     quality: str = "medium"