提交

kongdeqiang

2026-03-13 28ccfbfc51068a663a80764e14074df5edf2b5ba

 funasr/datasets/audio_datasets/index_ds.py

@@ -35,7 +35,7 @@
            with open(path, encoding="utf-8") as fin:
                file_list_all = fin.readlines()

                num_per_slice = (len(file_list_all) - 1) // data_split_num + 1
                num_per_slice = (len(file_list_all) - 1) // data_split_num + 1  # 16
                file_list = file_list_all[
                    data_split_i * num_per_slice : (data_split_i + 1) * num_per_slice
                ]
@@ -104,10 +104,10 @@
                            or target_len > self.max_target_length
                        ):
                            continue
                            

                        if (source_len + target_len) > self.max_token_length:
                            continue
                            

                        contents_i = {
                            "source": source,
                            "prompt": prompt,
@@ -118,6 +118,12 @@
                        text_language = data.get("text_language", None)
                        if text_language is not None:
                            contents_i["text_language"] = text_language
                        if "emo_target" in data:
                            contents_i["emo_target"] = data["emo_target"]
                        if "event_target" in data:
                            contents_i["event_target"] = data["event_target"]
                        if "with_or_wo_itn" in data:
                            contents_i["with_or_wo_itn"] = data["with_or_wo_itn"]
                        # audio_language = data.get("audio_language", None)
                        # if audio_language is not None:
                        #     contents_i["audio_language"] = audio_language