Thanks for another great video. Do you think you could use dual PDM microphones in a background or far field noise cancelling application? One microphone near the mouth and one further away. Convert both inputs to digital (I2S?) and subtract the far signal from the near signal? Or would you use a different way? I'm thinking about removing cabin noise in a vehicle. Thanks again!
No need to locate these 2 Microphones as you said. Usually dual-MIC NS algorithm (BF is probably used) requires standard location of MICs. For cabin noise, you can refer some successful cases of algorithm companies or vehicle manufactures with MIC arrays.
Should it work if I just run code on my board (I'm using F411E discovery) and connect my headphones to it? Well, it isn't, but surely I did something wrong. Besides it's not clear for me why are you using such sizes for the buffers and why they differ: 128, 256 and 16. Also I have no idea why only i'th and (i+2)'th samples are transmitted. How is it connected with the fact that you use 24b data of 32b frame and "half-word" configuration in DMA? I also wonder what is the reason of using slave mode, wouldn't it work if both I2S were set to master mode? Sorry, I am just a beginner and maybe I just didn't understand but I would be very grateful to hear some more thorough explaination on these details.
Hello, have you managed to figure out how to do this on F411E? I watched the video which was replied by. But do I really need to manually install drivers now?
@@theprotector11 yes. otherwise the DAC chip won't play the audio on the headphones. I have managed to work it on DISCO board .I can share you the project
@@theprotector11 I will share the github link. you will need a stm32f4 disco board. The board has cs432l audio chip and a pdm microphone. The driver for audio chip is only two files .c and .h which you need to include in this video's source code . Besides the code you will need to make two wired connections between I2S master and slave shown in this video.
Congratulations, great explanation. If I wanted to transfer the PDM data acquired by the microphones via USB_OTG_FS what should I activate? Which library do you use for data transfer? Thanks so much
Hi, I'm pretty new to DSP but how do you output the PCM data to say a waveform output like Audacity/Adobe Audition? Or even how do you output your real-time voice to something I can view off?
Very good explanation thx, I wonder if you could make another video about the same PDM bitstream but outputting it through the USB of the discovery board as a USB microphone application, that will be an awesome useful application.
Thanks so much for your videos! only one question, can i use this code with the DAC of the discovery board? What I need to change to make it work with it? Thans!
I am using an STM32F407VG. I downloaded the code to check if my board mic works. The code did not do anything. Can you confirm the hardware and software requirements of this project?
It's because he is using external codec, to use internal codec (audio jack), use the MY_CS43L22 library from th-cam.com/video/QIPQOnVablY/w-d-xo.html&ab_channel=MutexEmbedded-Education. If you use the same code from YetAnotherElectornicsChannel it will work just by starting the internal codec in I2S Mode. (remember to make the two wire connections mentiones it the video (clock from i2s3 to i2s2))
Hey, I have a question, why do you have "Hal_SAI_receive.... ,64);" instead of "Hal_SAI_receive...,128);", considering that pdmRxBuf size is 128 ? With your code, doesnt pdmRxBuf(64->127) all equal to 0?
This is a great question. I see by debugging that the second half of the pdmRxBuf is not zero, but I don't understand why. I too think I would normally start DMA with a length of 128, yet this causes the program to misbehave badly!
Hey man.. thanks again for all your work.. is it possible to use only one I2S interface.. like in previous videos.. in full duplex mode .. to use the mic and output the signal ?
@@YetAnotherElectronicsChannel I am using wolfson audio card with sm32f407vg. How can I know whether my board i working or not. s there any code for this? bro. Thanks
Also connected PDM microphone. There is the white like noise. With analog micrphones its ok it has clear sound. And in your video the PDM to PCM sound is not high quality. Is it acceptable for MEMS PDM microphones with PCM conversion or there is a methods to improve the quality?
Hello I'm building an audio circuit with the built-in PDM microphone and the 24-bit DAC CS43L22. So far i've only manage to build one via USB-Host. How can i build a similar demo such as yours? Thank you so much and love your video it was very helpful :)
sorry for my confusion.. the output sound is going to the integrated DAC right? if so.. is the DAC connected to the black headphone jac? how are you getting the sound from the DAC ?
For all sort of processing activities where you have a continuous data flow (and audio is such an application use-case) you are usually using double-buffers. Most DMA engines of the different uC's are supporting that usually. The idea is: You always have one buffer where you are the processing on with your processor-core and the other buffer is used to transmit/receive. As soon as the tx/rx operation is finished on one buffer, it will start rx/tx on the other buffer and you can start processing on the first buffer. This is doing then a ping-pong back and forth between those two buffers. Here in this example, you have one big array of data which is "virtually" divided in two buffers. The Half- and Full- complete callback are then indicating which buffer is ready to process.
What changes do you have to make to the configuration to make it work for two channels (two PDM microphones)? I assume you have to set in_ptr_channels = 2. Are there any other changes you would need to make to the configuration?
Hello! Bro I really appriciate and enjoy your videos. But I'm wondering about fifo pointers in your source code.(like fifobuf_w_ptr, fifobuf_r_ptr). They are increasing every time the functions like FifoRead(), FifoRead() are called. Is it Ok?
I haven’t checked the code but I think they’re relying on the fact that they start again with 0 as soon as they overflow. For an 8 bit unsigned type then at 255
![[#22] Calculating IIR parameters - Audio DSP On STM32 (24 Bit / 48 kHz)](http://i.ytimg.com/vi/Dt07Iwh6evE/mqdefault.jpg)
![[#22] Calculating IIR parameters - Audio DSP On STM32 (24 Bit / 48 kHz)](/img/tr.png)
![[#23] FFT Spectrum Analysis - Audio DSP On STM32 (24 Bit / 48 kHz)](http://i.ytimg.com/vi/3WF4CGKoMas/mqdefault.jpg)
![[#5] IIR Filters - Audio DSP On STM32 with I2S (24 Bit / 96 kHz)](http://i.ytimg.com/vi/lNBrGOk0XzE/mqdefault.jpg)
These DSP videos have helped me to make a very configurable ultrasonic leak detector. Fantastic quick-start.
Great video!! If anyone is wondering how to enable PDM2PCM option then make sure that CRC is activated.
I have a question. Why MidBuffer has only 16 elements if you process 64 elements of pdmRxBuf in PDM_Filter function?
Thanks for another great video. Do you think you could use dual PDM microphones in a background or far field noise cancelling application? One microphone near the mouth and one further away. Convert both inputs to digital (I2S?) and subtract the far signal from the near signal? Or would you use a different way? I'm thinking about removing cabin noise in a vehicle. Thanks again!
No need to locate these 2 Microphones as you said. Usually dual-MIC NS algorithm (BF is probably used) requires standard location of MICs. For cabin noise, you can refer some successful cases of algorithm companies or vehicle manufactures with MIC arrays.
Should it work if I just run code on my board (I'm using F411E discovery) and connect my headphones to it? Well, it isn't, but surely I did something wrong. Besides it's not clear for me why are you using such sizes for the buffers and why they differ: 128, 256 and 16. Also I have no idea why only i'th and (i+2)'th samples are transmitted. How is it connected with the fact that you use 24b data of 32b frame and "half-word" configuration in DMA? I also wonder what is the reason of using slave mode, wouldn't it work if both I2S were set to master mode? Sorry, I am just a beginner and maybe I just didn't understand but I would be very grateful to hear some more thorough explaination on these details.
You need to use drivers for the onboard audio DAC chip . Example th-cam.com/video/QIPQOnVablY/w-d-xo.html
Hello, have you managed to figure out how to do this on F411E? I watched the video which was replied by. But do I really need to manually install drivers now?
@@theprotector11 yes. otherwise the DAC chip won't play the audio on the headphones. I have managed to work it on DISCO board .I can share you the project
@@kishoresaldanha815 Yes please. How exactly can we get in contact here? And do I need the driver as well for the microphone to send signal as well?
@@theprotector11 I will share the github link. you will need a stm32f4 disco board. The board has cs432l audio chip and a pdm microphone. The driver for audio chip is only two files .c and .h which you need to include in this video's source code . Besides the code you will need to make two wired connections between I2S master and slave shown in this video.
It's an amazing video brother just wanted to see the DAC setup(Entire Hardware setup)
Congratulations, great explanation. If I wanted to transfer the PDM data acquired by the microphones via USB_OTG_FS what should I activate? Which library do you use for data transfer? Thanks so much
Hi, I'm pretty new to DSP but how do you output the PCM data to say a waveform output like Audacity/Adobe Audition? Or even how do you output your real-time voice to something I can view off?
No demo? So did it actually work?
Very good explanation thx, I wonder if you could make another video about the same PDM bitstream but outputting it through the USB of the discovery board as a USB microphone application, that will be an awesome useful application.
Can you create a tutorial for preamp for microphone with the dsp?
Thanks so much for your videos! only one question, can i use this code with the DAC of the discovery board? What I need to change to make it work with it? Thans!
No this demo code only works with the external PMOD audio board. I didn’t try to use it with the onboard DAC. Sorry
I am using an STM32F407VG. I downloaded the code to check if my board mic works. The code did not do anything. Can you confirm the hardware and software requirements of this project?
same here!
i think he used an external codec not the built in one on the board
It's because he is using external codec, to use internal codec (audio jack), use the MY_CS43L22 library from th-cam.com/video/QIPQOnVablY/w-d-xo.html&ab_channel=MutexEmbedded-Education. If you use the same code from YetAnotherElectornicsChannel it will work just by starting the internal codec in I2S Mode. (remember to make the two wire connections mentiones it the video (clock from i2s3 to i2s2))
Yet another cool video ! Thanks for sharing your knowledge :)
Great video, how would you send the PCM data over Uart?
Hey, I have a question, why do you have "Hal_SAI_receive.... ,64);" instead of "Hal_SAI_receive...,128);", considering that pdmRxBuf size is 128 ? With your code, doesnt pdmRxBuf(64->127) all equal to 0?
This is a great question. I see by debugging that the second half of the pdmRxBuf is not zero, but I don't understand why. I too think I would normally start DMA with a length of 128, yet this causes the program to misbehave badly!
Hey man.. thanks again for all your work.. is it possible to use only one I2S interface.. like in previous videos.. in full duplex mode .. to use the mic and output the signal ?
Hey, there are some pins denoted as GPIO_Output? How did you decide them? Can you please explain?
Thx for your valuable response.
show the hardware connection also
Please check 5:52. There it is shown
@@YetAnotherElectronicsChannel I am using wolfson audio card with sm32f407vg. How can I know whether my board i working or not. s there any code for this? bro.
Thanks
You guy is amazing. I lerned a lot from you.
thanks for your time and your knowledge!
Also connected PDM microphone. There is the white like noise. With analog micrphones its ok it has clear sound. And in your video the PDM to PCM sound is not high quality. Is it acceptable for MEMS PDM microphones with PCM conversion or there is a methods to improve the quality?
Hello I'm building an audio circuit with the built-in PDM microphone and the 24-bit DAC CS43L22. So far i've only manage to build one via USB-Host. How can i build a similar demo such as yours? Thank you so much and love your video it was very helpful :)
Congratulations, very good quality tutorial. Do you have time to make one where the PDM signal is fed to spectrum analyzer?
Perfect timing! Thanks a lot!
At 14:15 what is the guarantee that fifo_buf won't overflow ?
The input stream and output stream are derived from the same BCLK. So they are exactly in sync. So there won't be any buffer overflow / underflow
sorry for my confusion.. the output sound is going to the integrated DAC right? if so.. is the DAC connected to the black headphone jac? how are you getting the sound from the DAC ?
No I have used an external DAC converter (on the PMOD I2S2 Board). And that output I had recorded via the line inputs of my Zoom H4
@@YetAnotherElectronicsChannel thank you so much. Amazing videos.
Please can you explain to me further the role of HAL_I2S_TxHalfCpltCallback. Because I did not find a good explanation on the Net ?
For all sort of processing activities where you have a continuous data flow (and audio is such an application use-case) you are usually using double-buffers. Most DMA engines of the different uC's are supporting that usually.
The idea is: You always have one buffer where you are the processing on with your processor-core and the other buffer is used to transmit/receive.
As soon as the tx/rx operation is finished on one buffer, it will start rx/tx on the other buffer and you can start processing on the first buffer. This is doing then a ping-pong back and forth between those two buffers.
Here in this example, you have one big array of data which is "virtually" divided in two buffers. The Half- and Full- complete callback are then indicating which buffer is ready to process.
It's very kind of you to answer me.
If you don't mind, can i send you some questions by email where I found problems?
What changes do you have to make to the configuration to make it work for two channels (two PDM microphones)? I assume you have to set in_ptr_channels = 2. Are there any other changes you would need to make to the configuration?
Yes actually it should work somehow like that. Never tried it out to be honest
Great video, thanks for the good explanation!
Life saving ❤️️❤️️
Great Video! You Should implement some "chorus" effect on the STM32
At least I have some video about reverb effect: th-cam.com/video/nRLXNmLmHqM/w-d-xo.html
@@YetAnotherElectronicsChannel Yes, i already seen it.
Microphone beam forming????
NOT GETTING ANYTHING IN pdmRxBuf[]
Hello! Bro
I really appriciate and enjoy your videos.
But I'm wondering about fifo pointers in your source code.(like fifobuf_w_ptr, fifobuf_r_ptr).
They are increasing every time the functions like FifoRead(), FifoRead() are called. Is it Ok?
I haven’t checked the code but I think they’re relying on the fact that they start again with 0 as soon as they overflow. For an 8 bit unsigned type then at 255
@@YetAnotherElectronicsChannel Oh! I see. the fifo array size is 256 and the type of fifo_r_ptr and fifo_w_ptr are uint8_t, Thannks bro.
thanks so much for this very useful video
so useful
very good 👍
Спасибо
techRizz