diff --git a/.env.example b/.env.example
index e8e9079..16b6f50 100644
--- a/.env.example
+++ b/.env.example
@@ -42,10 +42,11 @@ OPENROUTER_API_KEY=sk-or-v1-your-key
 # LITELLM_API_KEY=sk-litellm-your-key
 # Admin can discover available models via the admin panel
 #
-# LiteLLM Speech-to-Text (auto-enabled when LITELLM_API_BASE is set)
+# LiteLLM Speech-to-Text
 # TRANSCRIBE_PROVIDER=litellm
-# LiteLLM STT: NOT supported for Vertex AI Chirp via LiteLLM proxy.
-# Use TRANSCRIBE_PROVIDER=google for Vertex AI / Gemini STT.
+# LITELLM_STT_MODEL=whisper-1             # model_name alias in LiteLLM config
+# NOTE: vertex_ai/chirp does NOT work via LiteLLM audio proxy.
+# For Vertex AI speech, use TRANSCRIBE_PROVIDER=google (Gemini inline audio).
 #
 # LiteLLM TTS: WORKS — use the model_name alias from your LiteLLM model_list
 # TTS_PROVIDER=litellm                  (auto-detected when LITELLM_API_BASE set)
diff --git a/src/routes/transcribe.js b/src/routes/transcribe.js
index f8a93f7..3e26994 100644
--- a/src/routes/transcribe.js
+++ b/src/routes/transcribe.js
@@ -78,7 +78,10 @@ router.post('/transcribe', authMiddleware, upload.single('audio'), async (req, r
 
     if (provider === 'litellm') {
       if (!process.env.LITELLM_API_BASE) return res.status(400).json({ error: 'LITELLM_API_BASE not set.' });
-      var sttModel = process.env.LITELLM_STT_MODEL || 'vertex_ai/chirp';
+      // Default to whisper-1 — vertex_ai/chirp does NOT work via LiteLLM audio proxy.
+      // User should set LITELLM_STT_MODEL to their configured whisper alias, or
+      // use TRANSCRIBE_PROVIDER=google for Vertex AI / Gemini transcription instead.
+      var sttModel = process.env.LITELLM_STT_MODEL || 'whisper-1';
       var FormData = require('form-data');
       var axios    = require('axios');
       var form = new FormData();