Add pad_or_trim function to handle segment before encoding (#705)

faster_whisper/transcribe.py

@@ -11,7 +11,7 @@ import ctranslate2
 import numpy as np
 import tokenizers
 
-from faster_whisper.audio import decode_audio
+from faster_whisper.audio import decode_audio, pad_or_trim
 from faster_whisper.feature_extractor import FeatureExtractor
 from faster_whisper.tokenizer import _LANGUAGE_CODES, Tokenizer
 from faster_whisper.utils import download_model, format_timestamp, get_end, get_logger
@@ -492,6 +492,7 @@ class WhisperModel:
             )
             segment = features[:, seek : seek + segment_size]
             segment_duration = segment_size * self.feature_extractor.time_per_frame
+            segment = pad_or_trim(segment, self.feature_extractor.nb_max_frames)
 
             if self.logger.isEnabledFor(logging.DEBUG):
                 self.logger.debug(