jerryyip

@raiscui 现有的respeakerd是不建议配合其他关键词引擎使用的，如果真要用的话有个别扭的方法可以试试：把`src/main.cc`里面的 `respeaker->RegisterOutputNode(vep_kws.get());`改成`respeaker->RegisterOutputNode(vep_bf.get());`，这样respeakerd输出的音频就是：做了beamforming的6个麦克风方向的音频+1路没做beamforming的音频+1路回采通道的音频，每一路都占8ms，128*size(int16)个字节。拿到各个方向的音频数据你就可以自己做kws了。 关于取消beamforming的问题，是无法取消的，因为manual_doa模式下，关键词引擎总是要监听一路音频数据。

Hi, Could you type "df -h" command and post what it returns on here? It seems respeaker ran out of its flash.

``` [ 464.020000] ****** wm8960_postinit ****** [ 464.040000] playback free_dma_buffer [ 464.050000] ptri2s_config->mmap_index:8 [ 464.060000] retire_capture_urb: 5 callbacks suppressed [ 473.680000] ummap MMAP[0]=0x81420000 [ 473.680000] playback free_dma_buffer [ 473.690000] ptri2s_config->mmap_index:8...

And it seems your "git" causes the problem above. So I advise you try this first: https://github.com/respeaker/get_started_with_respeaker/blob/master/QuickStart.md#first-impression-with-voice-interaction---respeaker-play-music

Hi, Have you try to mount mmcblk0p1 manually?

Hi @burton666 Please check your power adapter and see if it provides enough current to respeaker(preferably 1A)

Hi, I think you should try our hid tool on Windows, there are some bugs on Linux. And the normal operation is the leds are controlled by xmos on mic...

hi Sorry for my unclear answer. You could try as the picture shows.  |data|degree| |---|---| |1e, 0|30| |e, 1|270| |d2,0|210| |96,0|150|

And to test the mic array, 1Khz sound should be loud and close to mic array. Because the test program will run 10 times and there must be more than...

@claurier @joansc Will test this.