Upload KotobaWhisperPipeline

kotoba_whisper.py

@@ -216,7 +216,7 @@ class KotobaWhisperPipeline(AutomaticSpeechRecognitionPipeline):
                         item["is_last"] = m == len(timelines) - 1 and n == len(labels) - 1 and item["is_last"]
                         yield item
                 else:
-                    if inputs.shape[0] > self.feature_extractor.n_samples:
+                    if audio_array.shape[0] > self.feature_extractor.n_samples:
                         processed = self.feature_extractor(
                             audio_array,
                             sampling_rate=self.feature_extractor.sampling_rate,