fix: remote-queue-chat

[weedge]

fix:

• remote-queue-chat with redis queue
• session
• filter_special_chars for tts
• add is_save_record to save record and is_save_play_chunks to save play chat round record; default False

run:

• fe:

  
  TTS_TAG=tts_edge \
  REDIS_PASSWORD=***  SEND_KEY=*** \
  RUN_OP=fe \
  RECORDER_TAG=rms_recorder \
  python -m src.cmd.remote-queue-chat.generate_audio2audio > ./log/fe_std_out.log

or use wake word

TTS_TAG=tts_edge \ REDIS_PASSWORD= SEND_KEY= \ RUN_OP=fe \ RECORDER_TAG=wakeword_rms_recorder \ python -m src.cmd.remote-queue-chat.generate_audio2audio > ./log/fe_std_out.log

- be on cpu:

TQDM_DISABLE=True \ ASR_TAG=whisper_faster_asr \ ASR_MODEL_NAME_OR_PATH=./models/Systran/faster-whisper-base \ REDIS_PASSWORD= SEND_KEY= \ RUN_OP=be \ TTS_TAG=tts_edge \ LLM_MODEL_NAME=qwen \ LLM_MODEL_PATH=./models/qwen2-1_5b-instruct-q8_0.gguf \ python -m src.cmd.remote-queue-chat.generate_audio2audio > ./log/be_std_out.log

- be on gpu with local asr -> llm -> tts:

qwen1.5

ASR_TAG=whisper_faster_asr \ ASR_MODEL_NAME_OR_PATH=./models/Systran/faster-whisper-base \ REDIS_PASSWORD=$redis_pwd SEND_KEY=*** \ RUN_OP=be \ TQDM_DISABLE=True \ TTS_TAG=tts_coqui \ N_GPU_LAYERS=33 FLASH_ATTN=1 \ LLM_MODEL_NAME=qwen \ LLM_MODEL_PATH=./models/qwen1_5-7b-chat-q8_0.gguf \ python -m src.cmd.remote-queue-chat.generate_audio2audio > ./log/be_std_out.log

qwen2

TQDM_DISABLE=True \ ASR_TAG=whisper_faster_asr \ ASR_MODEL_NAME_OR_PATH=./models/Systran/faster-whisper-base \ REDIS_PASSWORD=$redis_pwd SEND_KEY=*** \ RUN_OP=be \ TTS_TAG=tts_coqui \ N_GPU_LAYERS=29 FLASH_ATTN=1 \ LLM_MODEL_NAME=qwen \ LLM_MODEL_PATH=./models/qwen2-1_5b-instruct-q8_0.gguf \ python -m src.cmd.remote-queue-chat.generate_audio2audio > ./log/be_std_out.log