Normalize transcription naming

AGENTS.md

@@ -4,7 +4,7 @@
 
 - `src/leld.py` is the primary entrypoint (X11 transcription daemon).
 - `src/recorder.py` handles audio capture using PortAudio via `sounddevice`.
-- `src/stt.py` wraps faster-whisper for transcription.
+- `src/transcription.py` wraps faster-whisper for transcription.
 
 ## Build, Test, and Development Commands
 

README.md

@@ -35,7 +35,7 @@ Create `~/.config/lel/config.json`:
 {
   "daemon": { "hotkey": "Cmd+m" },
   "recording": { "input": "0" },
-  "transcribing": { "model": "base", "device": "cpu" },
+  "transcription": { "model": "base", "device": "cpu" },
   "injection": { "backend": "clipboard" },
 
   "ai_cleanup": {

src/config.py

@@ -12,7 +12,7 @@ def _parse_bool(val: str) -> bool:
 class Config:
     daemon: dict = field(default_factory=lambda: {"hotkey": "Cmd+m"})
     recording: dict = field(default_factory=lambda: {"input": ""})
-    transcribing: dict = field(default_factory=lambda: {"model": "base", "device": "cpu"})
+    transcription: dict = field(default_factory=lambda: {"model": "base", "device": "cpu"})
     injection: dict = field(default_factory=lambda: {"backend": "clipboard"})
     ai_cleanup: dict = field(
         default_factory=lambda: {
@@ -36,15 +36,17 @@ def load(path: str | None) -> Config:
     p = Path(path) if path else default_path()
     if p.exists():
         data = json.loads(p.read_text(encoding="utf-8"))
-        if any(k in data for k in ("daemon", "recording", "transcribing", "injection", "ai_cleanup", "ai")):
+        if any(k in data for k in ("daemon", "recording", "transcription", "transcribing", "injection", "ai_cleanup", "ai")):
             for k, v in data.items():
                 if hasattr(cfg, k):
                     setattr(cfg, k, v)
+            if "transcribing" in data and "transcription" not in data:
+                cfg.transcription = data.get("transcribing", cfg.transcription)
         else:
             cfg.daemon["hotkey"] = data.get("hotkey", cfg.daemon["hotkey"])
             cfg.recording["input"] = data.get("ffmpeg_input", cfg.recording["input"])
-            cfg.transcribing["model"] = data.get("whisper_model", cfg.transcribing["model"])
+            cfg.transcription["model"] = data.get("whisper_model", cfg.transcription["model"])
-            cfg.transcribing["device"] = data.get("whisper_device", cfg.transcribing["device"])
+            cfg.transcription["device"] = data.get("whisper_device", cfg.transcription["device"])
             cfg.injection["backend"] = data.get("injection_backend", cfg.injection["backend"])
             cfg.ai_cleanup["enabled"] = data.get("ai_enabled", cfg.ai_cleanup["enabled"])
             cfg.ai_cleanup["model"] = data.get("ai_model", cfg.ai_cleanup["model"])
@@ -56,8 +58,8 @@ def load(path: str | None) -> Config:
         cfg.daemon = {"hotkey": "Cmd+m"}
     if not isinstance(cfg.recording, dict):
         cfg.recording = {"input": ""}
-    if not isinstance(cfg.transcribing, dict):
+    if not isinstance(cfg.transcription, dict):
-        cfg.transcribing = {"model": "base", "device": "cpu"}
+        cfg.transcription = {"model": "base", "device": "cpu"}
     if not isinstance(cfg.injection, dict):
         cfg.injection = {"backend": "clipboard"}
     if not isinstance(cfg.ai_cleanup, dict):
@@ -68,19 +70,21 @@ def load(path: str | None) -> Config:
             "base_url": "http://localhost:11434",
             "api_key": "",
         }
-    if isinstance(getattr(cfg, "ai", None), dict) and not cfg.ai_cleanup:
+    legacy_ai = getattr(cfg, "ai", None)
-        cfg.ai_cleanup = cfg.ai
+    if isinstance(legacy_ai, dict) and not cfg.ai_cleanup:
-    if hasattr(cfg, "ai"):
+        cfg.ai_cleanup = legacy_ai
     try:
         delattr(cfg, "ai")
+    except AttributeError:
+        pass
     except Exception:
         pass
 
     # env overrides
     if os.getenv("WHISPER_MODEL"):
-        cfg.transcribing["model"] = os.environ["WHISPER_MODEL"]
+        cfg.transcription["model"] = os.environ["WHISPER_MODEL"]
     if os.getenv("WHISPER_DEVICE"):
-        cfg.transcribing["device"] = os.environ["WHISPER_DEVICE"]
+        cfg.transcription["device"] = os.environ["WHISPER_DEVICE"]
     if os.getenv("WHISPER_FFMPEG_IN"):
         cfg.recording["input"] = os.environ["WHISPER_FFMPEG_IN"]
 

src/leld.py

@@ -11,7 +11,7 @@ from pathlib import Path
 
 from config import Config, load, redacted_dict
 from recorder import start_recording, stop_recording
-from stt import FasterWhisperSTT, STTConfig
+from transcription import FasterWhisperTranscriber, TranscriptionConfig
 from aiprocess import AIConfig, build_processor
 from inject import inject
 from x11_hotkey import listen
@@ -34,11 +34,11 @@ class Daemon:
         self.proc = None
         self.record = None
         self.timer = None
-        self.stt = FasterWhisperSTT(
+        self.transcriber = FasterWhisperTranscriber(
-            STTConfig(
+            TranscriptionConfig(
-                model=cfg.transcribing.get("model", "base"),
+                model=cfg.transcription.get("model", "base"),
                 language=None,
-                device=cfg.transcribing.get("device", "cpu"),
+                device=cfg.transcription.get("device", "cpu"),
                 vad_filter=True,
             )
         )
@@ -117,10 +117,10 @@ class Daemon:
 
         try:
             self.set_state(State.TRANSCRIBING)
-            logging.info("transcribing started")
+            logging.info("transcription started")
-            text = self.stt.transcribe(record.wav_path, language="en")
+            text = self.transcriber.transcribe(record.wav_path, language="en")
         except Exception as exc:
-            logging.error("stt failed: %s", exc)
+            logging.error("transcription failed: %s", exc)
             self.set_state(State.IDLE)
             return
 
@@ -129,7 +129,7 @@ class Daemon:
             self.set_state(State.IDLE)
             return
 
-        logging.info("stt: %s", text)
+        logging.info("transcription: %s", text)
 
         ai_enabled = self.cfg.ai_cleanup.get("enabled", False)
         ai_prompt_file = ""

src/transcription.py

@@ -6,7 +6,7 @@ from faster_whisper import WhisperModel
 
 
 @dataclass
-class STTConfig:
+class TranscriptionConfig:
     model: str
     language: str | None
     device: str
@@ -20,8 +20,8 @@ def _compute_type(device: str) -> str:
     return "int8"
 
 
-class FasterWhisperSTT:
+class FasterWhisperTranscriber:
-    def __init__(self, cfg: STTConfig):
+    def __init__(self, cfg: TranscriptionConfig):
         self.cfg = cfg
         self._model: WhisperModel | None = None