Improved logger

whisper_streaming/cli.py

@@ -3,7 +3,7 @@
 import argparse
 import queue
 import sys
-from logging import DEBUG, INFO, StreamHandler, getLogger
+from logging import DEBUG, INFO, StreamHandler, basicConfig, getLogger
 from typing import Optional, Union
 
 import sounddevice as sd
@@ -94,13 +94,11 @@ def get_opts() -> argparse.Namespace:
 
 def main() -> None:
     opts = get_opts()
+    basicConfig(
+        level=DEBUG if opts.debug else INFO,
+        format="[%(asctime)s] %(module)s.%(funcName)s %(levelname)s -> %(message)s",
+    )
 
-    root_logger = getLogger()
-    root_logger.addHandler(StreamHandler(sys.stderr))
-    if opts.debug:
-        root_logger.setLevel(DEBUG)
-    else:
-        root_logger.setLevel(INFO)
     if opts.beam_size <= 0:
         opts.beam_size = None
     try:

whisper_streaming/transcriber.py

@@ -195,7 +195,7 @@ class WhisperStreamingTranscriber:
             )
             if chunk is not None:
                 yield chunk
-            self.timestamp += float(segment_duration * HOP_LENGTH / SAMPLE_RATE)
+            self.timestamp += duration
 
         if result.temperature > 0.5:
             # do not feed the prompt tokens if a high temperature was used
@@ -217,7 +217,7 @@ class WhisperStreamingTranscriber:
         seek: int = 0
         rest_start: Optional[int] = None
         while seek < mel.shape[-1]:
-            logger.debug(seek)
+            logger.debug(f"seek={seek}, timestamp={self.timestamp}")
             segment = (
                 pad_or_trim(mel[:, :, seek:], N_FRAMES)
                 .to(self.model.device)  # type: ignore