alexfwulff / awulff-pico-playground Goto Github PK

command RECURSIVE_FIND_FILE not a CMake command, flagged up as syntax error by VS code and in the terminal when running Cmake -G "Nmake Makefiles" ..., and in VS code REVURSIVE_FIND_FILE

I've been trying to adapt your DMA ADC capture routine to simultaneously capture 2 channels but both captures seem to give the same data regardless of channel. Do you have any idea how to do 2 simultaneous DMA ADC captures correctly? I've tried polling each channel but the same thing happens so I'm clearly not understanding something.
I can get 2 unique DMA channel numbers from dma_channel_get_default_config and I've assumed most up to and including channel_config_set_dreq can just be doubled up, i.e.

adc_init();
adc_set_clkdiv(CLOCK_DIV);
adc_gpio_init(ADC_GPIO_L);
adc_select_input(DMA_CHANNEL_L);
adc_gpio_init(ADC_GPIO_R);
adc_select_input(DMA_CHANNEL_R);
adc_fifo_setup(true, true, 1, false, true);
sleep_ms(1000);
dma_chan_l = dma_claim_unused_channel(true);
dma_chan_r = dma_claim_unused_channel(true);
dma_cfg_l = dma_channel_get_default_config(dma_chan_l);
dma_cfg_r = dma_channel_get_default_config(dma_chan_r);
channel_config_set_transfer_data_size(&dma_cfg_l, DMA_SIZE_8);
channel_config_set_transfer_data_size(&dma_cfg_r, DMA_SIZE_8);
channel_config_set_read_increment(&dma_cfg_l, false);
channel_config_set_read_increment(&dma_cfg_r, false);
channel_config_set_write_increment(&dma_cfg_l, true);
channel_config_set_write_increment(&dma_cfg_r, true);
channel_config_set_dreq(&dma_cfg_l, DREQ_ADC);
channel_config_set_dreq(&dma_cfg_r, DREQ_ADC);

Hi @AlexFWulff,
I read your tutorial and it's honestly one of the few that actually clears concept about the inferencing rather than just doing stuff in the code.
I am trying to use your model to design an inference. It will make a certain pin high when a particular keyword is spoken, rest of the hardware is same (mic and pico).
Can you please tell me how should I go with it, just make the pin high when the state is changed? Do I need to use the second core for that?
Thanks.

Hi @AlexFWulff,

I am designing two models using Edge Impluse. One has Google Speech Commands as dataset and other is for binary classification of fire alarms. I will be using MFCC and MFE respectively for feature extraction. Will your pico-light-voice tutorial work for it if I want to make inference for pico using MAX9814. Basically I want to ask do I just have to make the changes you mentioned in your EdgeImpulse blog, if I change the dataset and feature extraction techniques? Keeping in mind that Google Speech Command Dataset has a sampling rate of 16 KHz.
Thanks so much.

So at the end of your guide, you say I should be able to see the outputs from the pico over usb. but looking at /dev/ nothing new is coming up. Also Thonny is not seeing any devices either. How am I supposed to see the outputs of the pico?