- use native speech detector

src/bt_signal.py

@@ -3,33 +3,22 @@
 import sys
 import os
 import pathlib
+from utils_types import SignalBoundaries
+from utils_sevana import speech_detector
 
 from pydub import silence, AudioSegment
 
-class SignalBoundaries:
-    # Offset from start (in seconds)
-    offset_start: float
-
-    # Offset from finish (in seconds)
-    offset_finish: float
-
-    def __init__(self, offset_start = 0.0, offset_finish = 0.0) -> None:
-        self.offset_start = offset_start
-        self.offset_finish = offset_finish
-
-    def __repr__(self) -> str:
-        return f'[offset_start: {round(self.offset_start, 3)}, offset_finish : {round(self.offset_finish, 3)}]'
-
+SILENCE_DELTA = 16
 
 def find_reference_signal(input_file: pathlib.Path, output_file: pathlib.Path = None, use_end_offset: bool = True) -> SignalBoundaries:
     myaudio = AudioSegment.from_wav(str(input_file))
     dBFS = myaudio.dBFS
     
     # Find silence intervals
-    intervals = silence.detect_nonsilent(myaudio, min_silence_len=1000, silence_thresh=dBFS-17, seek_step=50)
+    intervals = silence.detect_nonsilent(myaudio, min_silence_len=1000, silence_thresh=dBFS-SILENCE_DELTA, seek_step=50)
     
     # Translate to seconds
-    intervals = [((start/1000),(stop/1000)) for start,stop in intervals] #in sec
+    intervals = [((start/1000),(stop/1000)) for start,stop in intervals] # in sec
 
     # print(intervals)
 
@@ -48,6 +37,12 @@ def find_reference_signal(input_file: pathlib.Path, output_file: pathlib.Path =
 
     return SignalBoundaries()
 
+
+def find_reference_signal_via_speechdetector(input_file: pathlib.Path) -> SignalBoundaries:
+    bounds = speech_detector(str(input_file))
+    r = SignalBoundaries(bounds[0], bounds[1])
+    return bounds
+
 if __name__ == '__main__':
     if len(sys.argv) < 2:
         print(f'Please specify input filename.')