init

pipeline/kotoba_whisper.py

@@ -263,8 +263,8 @@ class KotobaWhisperPipeline(AutomaticSpeechRecognitionPipeline):
             outputs["chunks"] += new_chunk
         outputs["speaker_ids"] = sorted(set([o["speaker_id"] for o in outputs["chunks"]]))
         for s in outputs["speaker_ids"]:
-            outputs[f"chunk/{s}"] = sorted([o for o in outputs["chunks"] if o["speaker_id"] == s], key=lambda x: x["timestamp"][0])
-            outputs[f"text/{s}"] = "".join([i["text"] for i in outputs[f"chunk/{s}"]])
+            outputs[f"chunks/{s}"] = sorted([o for o in outputs["chunks"] if o["speaker_id"] == s], key=lambda x: x["timestamp"][0])
+            outputs[f"text/{s}"] = "".join([i["text"] for i in outputs[f"chunks/{s}"]])
             if postprocess_parameters["add_punctuation"]:
                 outputs[f"text/{s}"] = self.punctuator.punctuate(outputs[f"text/{s}"])
         return outputs