Problem in duplicating the training process

[angusxu458]

i am duplicating the training process proposed in your paper, and i'm new to AEC.

when i have

• *_doubletalk_lpb/mic.wav
• *_doubletalk_with_movement_lpb/mic.wav
• *_farend_singletalk_lpb/mic.wav
• *_farend_singletalk_with_movement_lpb/mic.wav
• *_nearend_singletalk_mic.wav
• *_sweep_lpb/mic.wav

which two should be the inputs of the model, and which should be the label. Thanks.

[angusxu458]

according your samples, i guess _lpb.wav and _mic.wav are the inputs, but which is the label among the above .wav? @breizhn

[angusxu458]

i have figured out the label should be *_nearend_singletalk_mic.wav

[angusxu458]

sorry i still have some problem for the input data (y,x, label),
i guess it should be like in the form as follow for the dataset AEC-Challenge/datasets/real/

(*_doubletalk_mic.wav, *_doubletalk_lpb.wav, *_nearend_singletalk_mic.wav)
(*_doubletalk_with_movement_mic.wav, *_doubletalk_with_movement_lpb.wav, *_nearend_singletalk_mic.wav)
(*_farend_singletalk_mic.wav, *_farend_singletalk_lpb.wav, zero_array)  # for no nearend speech
(*_farend_singletalk_with_movement_mic.wav, *_farend_singletalk_with_movement_lpb.wav, zero_array)

is there something wrong? can anyone help me!:pray:

[LXP-Never]

抱歉，我仍然对输入数据（y，x，标签）存在一些问题，
我想它应该像下面的数据集形式一样AEC-Challenge/datasets/real/

(*_doubletalk_mic.wav, *_doubletalk_lpb.wav, *_nearend_singletalk_mic.wav)
(*_doubletalk_with_movement_mic.wav, *_doubletalk_with_movement_lpb.wav, *_nearend_singletalk_mic.wav)
(*_farend_singletalk_mic.wav, *_farend_singletalk_lpb.wav, zero_array)  # for no nearend speech
(*_farend_singletalk_with_movement_mic.wav, *_farend_singletalk_with_movement_lpb.wav, zero_array)

有什么不对？谁能帮我！🙏

我也有同样的疑问，并且lpb代表什么我不知道

[LXP-Never]

sorry i still have some problem for the input data (y,x, label),
i guess it should be like in the form as follow for the dataset AEC-Challenge/datasets/real/

(*_doubletalk_mic.wav, *_doubletalk_lpb.wav, *_nearend_singletalk_mic.wav)
(*_doubletalk_with_movement_mic.wav, *_doubletalk_with_movement_lpb.wav, *_nearend_singletalk_mic.wav)
(*_farend_singletalk_mic.wav, *_farend_singletalk_lpb.wav, zero_array)  # for no nearend speech
(*_farend_singletalk_with_movement_mic.wav, *_farend_singletalk_with_movement_lpb.wav, zero_array)

is there something wrong? can anyone help me!🙏

大家都是中国人，我就用中文回复了，对于回声消除系统，通常使用两个输入信号，即输入近端麦克风信号和远端麦克风信号，输出近端语音信号。回声消除主要包括三个情况：
近端单讲：远端语音为静音，近端麦克风语音为***_nearend_singletalk_mic.wav，输出也是近端语音***_nearend_singletalk_mic.wav
远端单讲：远端语音为***_farend_singletalk_lpb，近端麦克风语音为远端语音的回声***_farend_singletalk_mic，输出静音
双讲：双讲情况我也没有弄懂，希望能和你交流一下，我的一个猜测是远端语音是***_doubletalk_lpb，近端麦克风语音***_doubletalk_mic，本来是要输出近端语音的，但是对应的近端语音数据集没有给到。你知道吗？求解答

[angusxu458]

sorry i still have some problem for the input data (y,x, label),
i guess it should be like in the form as follow for the dataset AEC-Challenge/datasets/real/

(*_doubletalk_mic.wav, *_doubletalk_lpb.wav, *_nearend_singletalk_mic.wav)
(*_doubletalk_with_movement_mic.wav, *_doubletalk_with_movement_lpb.wav, *_nearend_singletalk_mic.wav)
(*_farend_singletalk_mic.wav, *_farend_singletalk_lpb.wav, zero_array)  # for no nearend speech
(*_farend_singletalk_with_movement_mic.wav, *_farend_singletalk_with_movement_lpb.wav, zero_array)

is there something wrong? can anyone help me!🙏

大家都是中国人，我就用中文回复了，对于回声消除系统，通常使用两个输入信号，即输入近端麦克风信号和远端麦克风信号，输出近端语音信号。回声消除主要包括三个情况：
近端单讲：远端语音为静音，近端麦克风语音为***_nearend_singletalk_mic.wav，输出也是近端语音***_nearend_singletalk_mic.wav
远端单讲：远端语音为***_farend_singletalk_lpb，近端麦克风语音为远端语音的回声***_farend_singletalk_mic，输出静音
双讲：双讲情况我也没有弄懂，希望能和你交流一下，我的一个猜测是远端语音是***_doubletalk_lpb，近端麦克风语音***_doubletalk_mic，本来是要输出近端语音的，但是对应的近端语音数据集没有给到。你知道吗？求解答

这个数据集里应该没有

[LXP-Never]

sorry i still have some problem for the input data (y,x, label),
i guess it should be like in the form as follow for the dataset AEC-Challenge/datasets/real/

(*_doubletalk_mic.wav, *_doubletalk_lpb.wav, *_nearend_singletalk_mic.wav)
(*_doubletalk_with_movement_mic.wav, *_doubletalk_with_movement_lpb.wav, *_nearend_singletalk_mic.wav)
(*_farend_singletalk_mic.wav, *_farend_singletalk_lpb.wav, zero_array)  # for no nearend speech
(*_farend_singletalk_with_movement_mic.wav, *_farend_singletalk_with_movement_lpb.wav, zero_array)

is there something wrong? can anyone help me!🙏

大家都是中国人，我就用中文回复了，对于回声消除系统，通常使用两个输入信号，即输入近端麦克风信号和远端麦克风信号，输出近端语音信号。回声消除主要包括三个情况：
近端单讲：远端语音为静音，近端麦克风语音为***_nearend_singletalk_mic.wav，输出也是近端语音***_nearend_singletalk_mic.wav
远端单讲：远端语音为***_farend_singletalk_lpb，近端麦克风语音为远端语音的回声***_farend_singletalk_mic，输出静音
双讲：双讲情况我也没有弄懂，希望能和你交流一下，我的一个猜测是远端语音是***_doubletalk_lpb，近端麦克风语音***_doubletalk_mic，本来是要输出近端语音的，但是对应的近端语音数据集没有给到。你知道吗？求解答

这个数据集里应该没有

那该如何进行双讲回声消除呢？

[JuneRen]

sorry i still have some problem for the input data (y,x, label),
i guess it should be like in the form as follow for the dataset AEC-Challenge/datasets/real/

(*_doubletalk_mic.wav, *_doubletalk_lpb.wav, *_nearend_singletalk_mic.wav)
(*_doubletalk_with_movement_mic.wav, *_doubletalk_with_movement_lpb.wav, *_nearend_singletalk_mic.wav)
(*_farend_singletalk_mic.wav, *_farend_singletalk_lpb.wav, zero_array)  # for no nearend speech
(*_farend_singletalk_with_movement_mic.wav, *_farend_singletalk_with_movement_lpb.wav, zero_array)

is there something wrong? can anyone help me!pray

大家都是中国人，我就用中文回复了，对于回声消除系统，通常使用两个输入信号，即输入近端麦克风信号和远端麦克风信号，输出近端语音信号。回声消除主要包括三个情况：
近端单讲：远端语音为静音，近端麦克风语音为***_nearend_singletalk_mic.wav，输出也是近端语音***_nearend_singletalk_mic.wav
远端单讲：远端语音为***_farend_singletalk_lpb，近端麦克风语音为远端语音的回声***_farend_singletalk_mic，输出静音
双讲：双讲情况我也没有弄懂，希望能和你交流一下，我的一个猜测是远端语音是***_doubletalk_lpb，近端麦克风语音***_doubletalk_mic，本来是要输出近端语音的，但是对应的近端语音数据集没有给到。你知道吗？求解答

您好，我们最近也在研究这个数据集，请问您找到近端语音了吗？感谢分享下，谢谢~

[LXP-Never]

sorry i still have some problem for the input data (y,x, label),
i guess it should be like in the form as follow for the dataset AEC-Challenge/datasets/real/

(*_doubletalk_mic.wav, *_doubletalk_lpb.wav, *_nearend_singletalk_mic.wav)
(*_doubletalk_with_movement_mic.wav, *_doubletalk_with_movement_lpb.wav, *_nearend_singletalk_mic.wav)
(*_farend_singletalk_mic.wav, *_farend_singletalk_lpb.wav, zero_array)  # for no nearend speech
(*_farend_singletalk_with_movement_mic.wav, *_farend_singletalk_with_movement_lpb.wav, zero_array)

is there something wrong? can anyone help me!pray

大家都是中国人，我就用中文回复了，对于回声消除系统，通常使用两个输入信号，即输入近端麦克风信号和远端麦克风信号，输出近端语音信号。回声消除主要包括三个情况：
近端单讲：远端语音为静音，近端麦克风语音为***_nearend_singletalk_mic.wav，输出也是近端语音***_nearend_singletalk_mic.wav
远端单讲：远端语音为***_farend_singletalk_lpb，近端麦克风语音为远端语音的回声***_farend_singletalk_mic，输出静音
双讲：双讲情况我也没有弄懂，希望能和你交流一下，我的一个猜测是远端语音是***_doubletalk_lpb，近端麦克风语音***_doubletalk_mic，本来是要输出近端语音的，但是对应的近端语音数据集没有给到。你知道吗？求解答

您好，我们最近也在研究这个数据集，请问您找到近端语音了吗？感谢分享下，谢谢~

这个数据集没有提供近端语音，还是乖乖使用合成的吧

[JuneRen]

sorry i still have some problem for the input data (y,x, label),
i guess it should be like in the form as follow for the dataset AEC-Challenge/datasets/real/

(*_doubletalk_mic.wav, *_doubletalk_lpb.wav, *_nearend_singletalk_mic.wav)
(*_doubletalk_with_movement_mic.wav, *_doubletalk_with_movement_lpb.wav, *_nearend_singletalk_mic.wav)
(*_farend_singletalk_mic.wav, *_farend_singletalk_lpb.wav, zero_array)  # for no nearend speech
(*_farend_singletalk_with_movement_mic.wav, *_farend_singletalk_with_movement_lpb.wav, zero_array)

is there something wrong? can anyone help me!pray

大家都是中国人，我就用中文回复了，对于回声消除系统，通常使用两个输入信号，即输入近端麦克风信号和远端麦克风信号，输出近端语音信号。回声消除主要包括三个情况：
近端单讲：远端语音为静音，近端麦克风语音为***_nearend_singletalk_mic.wav，输出也是近端语音***_nearend_singletalk_mic.wav
远端单讲：远端语音为***_farend_singletalk_lpb，近端麦克风语音为远端语音的回声***_farend_singletalk_mic，输出静音
双讲：双讲情况我也没有弄懂，希望能和你交流一下，我的一个猜测是远端语音是***_doubletalk_lpb，近端麦克风语音***_doubletalk_mic，本来是要输出近端语音的，但是对应的近端语音数据集没有给到。你知道吗？求解答

您好，我们最近也在研究这个数据集，请问您找到近端语音了吗？感谢分享下，谢谢~

这个数据集没有提供近端语音，还是乖乖使用合成的吧

好吧，感谢分享~

[LXP-Never]

sorry i still have some problem for the input data (y,x, label),
i guess it should be like in the form as follow for the dataset AEC-Challenge/datasets/real/

(*_doubletalk_mic.wav, *_doubletalk_lpb.wav, *_nearend_singletalk_mic.wav)
(*_doubletalk_with_movement_mic.wav, *_doubletalk_with_movement_lpb.wav, *_nearend_singletalk_mic.wav)
(*_farend_singletalk_mic.wav, *_farend_singletalk_lpb.wav, zero_array)  # for no nearend speech
(*_farend_singletalk_with_movement_mic.wav, *_farend_singletalk_with_movement_lpb.wav, zero_array)

is there something wrong? can anyone help me!🙏

大家都是中国人，我就用中文回复了，对于回声消除系统，通常使用两个输入信号，即输入近端麦克风信号和远端麦克风信号，输出近端语音信号。回声消除主要包括三个情况： 近端单讲：远端语音为静音，近端麦克风语音为***_nearend_singletalk_mic.wav，输出也是近端语音***_nearend_singletalk_mic.wav 远端单讲：远端语音为***_farend_singletalk_lpb，近端麦克风语音为远端语音的回声***_farend_singletalk_mic，输出静音 双讲：双讲情况我也没有弄懂，希望能和你交流一下，我的一个猜测是远端语音是***_doubletalk_lpb，近端麦克风语音***_doubletalk_mic，本来是要输出近端语音的，但是对应的近端语音数据集没有给到。你知道吗？求解答

论文里面说了吧，训练只用远端参考信号和近端的回声信号，只考虑远端单讲的case

只用单讲可以