use fun-asr-nano model instead

README.md

@@ -69,8 +69,8 @@ The checked-in `docker-compose.yml` mounts local `./config.toml` there for the A
 ## Services And Ports
 
 - Frontend: `http://localhost:3000`
-- Backend API: `http://localhost:8080`
-- API docs: `http://localhost:8080/docs`
+- Backend API: `http://localhost:8081`
+- API docs: `http://localhost:8081/docs`
 - Redis: `localhost:6379` when exposed by Compose
 - PostgreSQL: only started by Compose when using the `local-db` profile
 
@@ -145,7 +145,7 @@ The frontend reads the backend URL from `frontend/public/config.js` at runtime:
 
 ```js
 window.__EVANESCERE_FRONTEND_CONFIG__ = {
-  apiBaseUrl: "http://192.168.1.44:8080"
+  apiBaseUrl: "http://192.168.1.44:8081"
 };
 ```
 

docker-compose.yml

@@ -10,7 +10,7 @@ services:
     build: .
     command: uvicorn evanescere.api:app --host 0.0.0.0 --port 8000
     ports:
-      - "8080:8000"
+      - "8081:8000"
     volumes:
       - ./config.toml:/etc/evanescere/config.toml:ro
       - ./storage:/data/evanescere

frontend/public/config.js

@@ -1,3 +1,3 @@
 window.__EVANESCERE_FRONTEND_CONFIG__ = {
-  apiBaseUrl: "http://192.168.1.44:8080"
+  apiBaseUrl: "http://192.168.1.44:8081"
 };

frontend/src/api.ts

@@ -9,7 +9,7 @@ import type {
 
 const runtimeApiBase = window.__EVANESCERE_FRONTEND_CONFIG__?.apiBaseUrl;
 
-export const apiBaseUrl = (runtimeApiBase || "http://localhost:8080").replace(
+export const apiBaseUrl = (runtimeApiBase || "http://localhost:8081").replace(
   /\/$/,
   "",
 );

src/evanescere/services/asr.py

@@ -35,7 +35,7 @@ class FunAsrClient:
                     "model": self.model,
                     "language": "zh",
                     "response_format": "verbose_json",
-                    "timestamp_granularities[]": "segment",
+                    "timestamp_granularities": "segment",
                 },
                 files={"file": (audio_path.name, audio_file, "audio/wav")},
                 timeout=None,
@@ -45,8 +45,12 @@ class FunAsrClient:
         logger.info("funasr request done audio=%s keys=%s", audio_path, sorted(payload.keys()))
         return payload
 
-
-def normalize_segments(payload: dict[str, Any]) -> list[dict[str, Any]]:
+def normalize_segments(
+    payload: dict[str, Any],
+    *,
+    fallback_start_sec: float = 0.0,
+    fallback_end_sec: float | None = None,
+) -> list[dict[str, Any]]:
     raw_segments = payload.get("segments") or payload.get("sentence_info") or []
     logger.debug("normalizing asr segments raw_count=%s", len(raw_segments))
     normalized: list[dict[str, Any]] = []
@@ -71,10 +75,17 @@ def normalize_segments(payload: dict[str, Any]) -> list[dict[str, Any]]:
             }
         )
     if not normalized and payload.get("text"):
+        logger.warning(
+            "asr response contained text but no timestamped segments; "
+            "falling back to the known audio bounds"
+        )
+        end_sec = fallback_end_sec
+        if end_sec is None:
+            end_sec = float(payload.get("duration") or fallback_start_sec + 0.1)
         normalized.append(
             {
-                "start_sec": 0.0,
-                "end_sec": 0.1,
+                "start_sec": fallback_start_sec,
+                "end_sec": max(fallback_start_sec + 0.1, end_sec),
                 "text": str(payload["text"]).strip(),
                 "speaker": None,
                 "confidence": None,

tests/test_asr.py

@@ -0,0 +1,16 @@
+from evanescere.services.asr import normalize_segments
+
+
+def test_normalize_segments_uses_known_audio_bounds_when_timestamps_are_missing():
+    assert normalize_segments(
+        {"text": "你好", "segments": []}, fallback_start_sec=0, fallback_end_sec=15
+    ) == [
+        {
+            "start_sec": 0,
+            "end_sec": 15,
+            "text": "你好",
+            "speaker": None,
+            "confidence": None,
+            "metadata": {"text": "你好", "segments": []},
+        }
+    ]