ZhangZhaofeng comments

Results 7 comments of


                                            ZhangZhaofeng

Can this code work on a tensorflow trained model

Thanks! Why I ask this question is because i run this code with a tensorflow trained LSTM model on CHiME3 task and decode with theano. The training is well done...

Can this code work on a tensorflow trained model

Thanks for your advice. The features are OK after `feat-to-len` checking each of them. I tried to use tensorflow backend with GPU decoding, the problem did not appear. memory can...

Can this code work on a tensorflow trained model

Seems that it is a theano problem. My theano version is 0.9.0. It will always cause some memory issues. After I downgrade to 0.8.2, the memory runs normally.

Some questions about mic_steering_vector

作者使用了一个圆阵，原理见我的回答 https://github.com/AkojimaSLP/Beamforming-for-speech-enhancement/issues/1#issuecomment-1480893214

Is linear array?

作者使用了一个圆阵，TDOA在计算时假设麦克风和中心点为由两个麦克风组成的线性阵列，那么TDOA的计算模型就可以简化为计算信号到达中心点的时间差。于是用下列方式计算这个时间差。 (self.mic_diameter / 2) * np.cos(np.deg2rad(look_direction) - np.deg2rad(mic_angle) / self.sound_speed 同时计算6个麦克风的时间差，即可得到6个目标麦克风的TDOA。

Is linear array?

> MIC_DIAMETER是麦克风的半径还是直径是直径。计算的时候要除以2。

灵活选择gpu

训练时可以选卡，推理的话好像不会自动选。如果能改config的话也行。我现在是在docker里面跑的，每次开始要把被占用的卡禁掉。