fix: add universal-2 fallback for languages unsupported by universal-… (#454)

universal-3-pro does not support all languages (e.g. Dutch/nl). Setting
speech_models to ["universal-3-pro", "universal-2"] lets AssemblyAI fall
back to universal-2 for the 99 languages not covered by universal-3-pro.

Also fetch and store the real AssemblyAI error reason in the webhook
error handler instead of the generic "AssemblyAI error for transcript
<id>".

Author: dtrn2048

echo/server/dembrane/api/webhooks.py

@@ -73,9 +73,15 @@ async def assemblyai_webhook_callback(
 
         if normalized_status == "error":
             from dembrane.tasks import _on_chunk_transcription_done
-            from dembrane.transcribe import _save_chunk_error
+            from dembrane.transcribe import _save_chunk_error, fetch_assemblyai_result
 
-            _save_chunk_error(chunk_id, f"AssemblyAI error for transcript {payload.transcript_id}")
+            error_detail = f"AssemblyAI error for transcript {payload.transcript_id}"
+            try:
+                fetch_assemblyai_result(payload.transcript_id)
+            except Exception as fetch_exc:
+                error_detail = str(fetch_exc)
+
+            _save_chunk_error(chunk_id, error_detail)
             _on_chunk_transcription_done(conversation_id, chunk_id, logger)
             delete_assemblyai_webhook_metadata(payload.transcript_id)
             return {"status": "error_handled"}

echo/server/dembrane/transcribe.py

@@ -107,7 +107,7 @@ def transcribe_audio_assemblyai(
 
     data: dict[str, Any] = {
         "audio_url": audio_file_uri,
-        "speech_models": ["universal-3-pro"],
+        "speech_models": ["universal-3-pro", "universal-2"],
         "language_detection": True,
         "language_detection_options": {
             "expected_languages": list(set(get_allowed_languages()) | {"pt"}),

echo/server/tests/test_transcribe_webhook.py

@@ -51,7 +51,7 @@ def _fake_post(url: str, **kwargs: Any) -> _FakeResponse:
     assert transcript is None
     assert payload == {"transcript_id": "tx-1"}
     assert captured["url"].endswith("/v2/transcript")
-    assert captured["json"]["speech_models"] == ["universal-3-pro"]
+    assert captured["json"]["speech_models"] == ["universal-3-pro", "universal-2"]
     assert "speech_model" not in captured["json"]
     assert "prompt" not in captured["json"]
     assert captured["json"]["keyterms_prompt"] == ["Dembrane"]
@@ -92,7 +92,7 @@ def _fake_get(_url: str, **_kwargs: Any) -> _FakeResponse:
     assert response["status"] == "completed"
     assert payloads["polls"] == 2
     post_payload = payloads["posts"][0]
-    assert post_payload["speech_models"] == ["universal-3-pro"]
+    assert post_payload["speech_models"] == ["universal-3-pro", "universal-2"]
     assert "speech_model" not in post_payload
     assert "webhook_url" not in post_payload