Restructuring the STT structure

apps/stt/.env.example

@@ -1,4 +1,6 @@
 # Local Pipecat Whisper — no GapGPT / OpenAI cloud key required for STT.
+# Use a local size: tiny | base | small | medium | large-v3
+# Do NOT use cloud names like whisper-1
 STT_MODEL=small
 STT_DEVICE=cpu
 STT_COMPUTE_TYPE=int8

apps/stt/main.py

@@ -39,8 +39,39 @@ def _require_token(header_token: str | None) -> None:
 
 
 def _model_name() -> str:
-    # faster-whisper / Pipecat: tiny | base | small | medium | large-v3 | ...
-    return _env("STT_MODEL", "small") or "small"
+    """Map env / legacy cloud names to Faster-Whisper local sizes."""
+    raw = (_env("STT_MODEL", "small") or "small").lower()
+    # Old GapGPT/OpenAI cloud id — not a local Whisper size
+    legacy = {
+        "whisper-1": "small",
+        "whisper": "small",
+        "gpt-4o-transcribe": "small",
+        "gpt-4o-mini-transcribe": "base",
+    }
+    if raw in legacy:
+        return legacy[raw]
+    allowed = {
+        "tiny",
+        "tiny.en",
+        "base",
+        "base.en",
+        "small",
+        "small.en",
+        "medium",
+        "medium.en",
+        "large",
+        "large-v1",
+        "large-v2",
+        "large-v3",
+        "distil-large-v2",
+        "distil-medium.en",
+        "distil-large-v3",
+        "large-v3-turbo",
+    }
+    if raw in allowed:
+        return raw
+    # unknown → safe default for Persian
+    return "small"
 
 
 def _stt_service() -> WhisperSTTService: