conf/data_t2s.yaml

TEXT2TOKEN_hq_add_2025_3_17:
  lab_path: "/home/node48_tmpdata/hkxie/4O/speech_data_final/10wh_token_data/TEXT2TOKEN_hq/shards_list.txt"
  path_huawei: "/mnt/sfs/asr/update_data/TEXT2TOKEN_hq_add_2025_3_17/shards_list.txt"
  path: /home/A03_tmpdata2/s2s/2000_hq_S2Chat_by_hkxie_added_by_20250411/combine_tts/combines_list.txt
  weight: 5

#english_text_token_add_2025_3_26:
#  path_huawei: "/mnt/sfs/asr/update_data/english_speech_data_final_TEXT2TOKEN_part_1_added_2025_3_26/shards_list.txt" # 2000
#  data_list_path: "/home/work_nfs14/code/hkxie/ASR/understanding_LLM_task/english/speech_data_final/data_libriheavy_part_1.list"
#  lab_path: "?"
#  weight: 0.5 #1 #1 #10
#english_TEXT2TOKEN_part_2_added_by_20250402:
#  path_huawei: "/mnt/sfs/asr/update_data/english_TEXT2TOKEN_part_2_added_by_20250402/shards_list.txt" #8050
#  data_list_path: "/home/work_nfs14/code/hkxie/ASR/understanding_LLM_task/english/speech_data_final/data_libriheavy_part_2.list"
#  lab_path: "?"
#  weight: 0.5 #1 #1 #10
#
#zh_en_mix_tts_added_by_20250402: # tts
#  path_huawei: "/mnt/sfs/asr/update_data/zh_en_mix_s2s_added_by_20250402/shards_list.txt" # 7
#  weight: 1
#poly_tts_added_by_20250402: # tts
#  path: "/mnt/sfs/asr/update_data/poly_s2s_added_by_20250402/shards_list.txt" # 295
#  weight: 0.5 #10
#
#text2token_itn_by_cywang_added_by_20250428: # zyzhang 负责，cywang打包
#  lab_path: "/home/work_nfs7/cywang/OSUM/OSUM_data/shard/text2token/tn/shards_list.txt"
#  path_huawei: "/mnt/sfs/asr/update_data/text2token_itn_by_cywang_added_by_20250428/shards_list.txt" # 1100
#  weight: 0.5