s2t 0.1.0.post1.dev2__py3-none-any.whl → 0.1.2__py3-none-any.whl

s2t/cli.py

@@ -256,6 +256,8 @@ def run_session(opts: SessionOptions) -> int:
     else:
         print("—" * 60)
         print("Transcript (clipboard text):")
+        # Visual separator before the actual transcript text
+        print("=" * 60)
         print(text_final.rstrip("\n"))
 
     if opts.profile:

s2t/whisper_engine.py

@@ -29,7 +29,10 @@ class WhisperEngine:
         self.samplerate = samplerate
         self.channels = channels
         self.verbose = verbose
-        self.profile = profile or {}
+        # Use the provided profile dict even if it's empty.
+        # Using `or {}` would create a new dict when an empty one is passed,
+        # breaking shared accumulation with the caller (CLI).
+        self.profile = profile if profile is not None else {}
         self._executor: ThreadPoolExecutor | None = None
 
     def preload(self) -> tuple[ThreadPoolExecutor | None, Future | None]:
@@ -75,10 +78,31 @@ class WhisperEngine:
         frames: int,
         initial_prompt: str | None = None,
     ) -> TranscriptionResult:
+        # Load audio without ffmpeg by reading via soundfile and passing a numpy array
+        # to Whisper. We ensure mono float32 at 16 kHz as expected by Whisper's API.
         task = "translate" if self.translate else "transcribe"
+        import numpy as np
+
+        try:
+            import soundfile as sf
+        except Exception as e:
+            raise RuntimeError("soundfile is required to read recorded audio.") from e
+
+        from .utils import resample_linear
+
+        # Read audio from file (supports WAV/FLAC via libsndfile), convert to mono
+        data, sr = sf.read(str(audio_path), dtype="float32", always_2d=True)
+        # data shape: (n_frames, n_channels). Convert to mono by averaging if needed
+        if data.ndim == 2 and data.shape[1] > 1:
+            mono = data.mean(axis=1)
+        else:
+            mono = data.reshape(-1)
+        # Resample to 16k expected by Whisper when passing arrays
+        mono_16k: np.ndarray = resample_linear(mono, int(sr), 16000)
+
         t0 = time.perf_counter()
         res: dict[str, Any] = model.transcribe(
-            str(audio_path),
+            mono_16k,
             task=task,
             language=self.language,
             fp16=False,

{s2t-0.1.0.post1.dev2.dist-info → s2t-0.1.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: s2t
-Version: 0.1.0.post1.dev2
+Version: 0.1.2
 Summary: Speech to Text (s2t): Record audio, run Whisper, export formats, and copy transcript to clipboard.
 Author: Maintainers
 License-Expression: LicenseRef-Proprietary

s2t-0.1.2.dist-info/RECORD

@@ -0,0 +1,14 @@
+s2t/__init__.py,sha256=wV4E9i-7KrUn1dOtLUQB3ZGEKx9gRWH3hPHlpw-ZdWc,332
+s2t/cli.py,sha256=_7zIhcdI7DI_3Dxs2EcvQkE-fSGclkJ2TjCvDYlI65E,15871
+s2t/config.py,sha256=mzz6ljGEupNDAzlUwf5kvl0iKqO8WZ4TWsU4nSVtp0M,409
+s2t/outputs.py,sha256=Lo8VcARZ7QPuuQQNu8myD5J4c4NO1Rs0L1DLnzLe9tM,1546
+s2t/py.typed,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
+s2t/recorder.py,sha256=uBD9mYf-uUCkRJw8fQitVnDrX6PwRNXJycyY4dBfXL0,8076
+s2t/types.py,sha256=BuMyWuueS7EZbk7I_CkIWSb69Yi6g9-wr7CZLAZKflw,242
+s2t/utils.py,sha256=YU6YhiuONmqhrKte4DY5tiC5PP-yFExJMMBzFUiA8qA,3416
+s2t/whisper_engine.py,sha256=Y7kTYnB-LtsVj2KvF60VcI37MNAVfmMcGSJMMxFyZgA,6220
+s2t-0.1.2.dist-info/METADATA,sha256=YPufi9Xb9aqCARVeRkzrDQ_b2UNi7Pkq1dJAJ7vkr68,4557
+s2t-0.1.2.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+s2t-0.1.2.dist-info/entry_points.txt,sha256=JISIUlZAJ3DX1dB6zT3X_E3vcXI-eWEQKwHiT35fPKs,37
+s2t-0.1.2.dist-info/top_level.txt,sha256=o8N0JcuHdIrfX3iGHvntHiDC2XgN7__joyNu08ZOh0s,4
+s2t-0.1.2.dist-info/RECORD,,

s2t-0.1.0.post1.dev2.dist-info/RECORD

@@ -1,14 +0,0 @@
-s2t/__init__.py,sha256=wV4E9i-7KrUn1dOtLUQB3ZGEKx9gRWH3hPHlpw-ZdWc,332
-s2t/cli.py,sha256=5Z0YxLPwvfV8wrU-vN1s1HzzOLmA0HYi5uVf6brUtQQ,15786
-s2t/config.py,sha256=mzz6ljGEupNDAzlUwf5kvl0iKqO8WZ4TWsU4nSVtp0M,409
-s2t/outputs.py,sha256=Lo8VcARZ7QPuuQQNu8myD5J4c4NO1Rs0L1DLnzLe9tM,1546
-s2t/py.typed,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
-s2t/recorder.py,sha256=uBD9mYf-uUCkRJw8fQitVnDrX6PwRNXJycyY4dBfXL0,8076
-s2t/types.py,sha256=BuMyWuueS7EZbk7I_CkIWSb69Yi6g9-wr7CZLAZKflw,242
-s2t/utils.py,sha256=YU6YhiuONmqhrKte4DY5tiC5PP-yFExJMMBzFUiA8qA,3416
-s2t/whisper_engine.py,sha256=s9NBPtyptdhKauKQB4moq2SeGDQp2z7qc13e8C00SxY,5075
-s2t-0.1.0.post1.dev2.dist-info/METADATA,sha256=c-7jrltbRiLjW0ixPZwgf49L8Ar7p7N5Dc7b0QO_pUo,4568
-s2t-0.1.0.post1.dev2.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-s2t-0.1.0.post1.dev2.dist-info/entry_points.txt,sha256=JISIUlZAJ3DX1dB6zT3X_E3vcXI-eWEQKwHiT35fPKs,37
-s2t-0.1.0.post1.dev2.dist-info/top_level.txt,sha256=o8N0JcuHdIrfX3iGHvntHiDC2XgN7__joyNu08ZOh0s,4
-s2t-0.1.0.post1.dev2.dist-info/RECORD,,