Having a blog post to replicate your result with the ReSpeaker 4 mic would be SUPER helpful. I already have the ReSpeaker 4 mic but found existing tutorials / code difficult to locate sound source.
Hello, I'm also working on an Acoular project and am trying to use audio samples that have worked to debug my code. Would really appreciate the audio files if possible!
Would be interesting to see what it shows when pointed at a stereo sound source. I.e. whether it will show that the sound is coming from two different speakers, or if it shows that the sound is coming from the “phantom” centre
I would really like to read another blog post about the ReSpeaker 4. While looking for the shield I also found the ReSpeaker 6-Mic. Did you try your Code with the ReSpeaker 4 and a custom MicGeom file as well? How big are the differences?
You create an H5 format file. Then, you merge a wav and avi file. At what step did you, and how, place the overlay on the video? The article seems to have a large gap at the end.
Hi Michael, this is a fantastic project. I'm wondering if you found the 4 mic array less sensitive to low frequency noise. I also wonder if this is possible with omnidirectional mics if their geometric spacing is known.
Hi Mike, Do you have git repository for the ReSpeaker and GCC Phat algorithm experiment? Would be great of you put it up and possibly others to contribute if they try it and maybe refine and add to it. Just a thought as I am interested to try it and contribute. Thanks Hoo
Can the current software capture specific frequencies? From reading your post, it seems the detection is based on the highest amplitude of the pressure wave. What if one is to introduce a tunable band-pass filter to help focus on specific frequencies only?
This is very interested! I am intending to do the same thing but in 3D (hemispherical shape). These are some of the hardware so far, please allow me to share
When I convert that to .h5 I assume it retains the same audio channels which results in an error when running the script to analyze the results.
---> 14 pm = bb.synthetic( 8000, 3 )
ValueError: 128 channels do not fit 16 mics
I'm sure this is something really simple to fix, but I'm not familiar enough to know how to alter that initial recording to only capture 16 channels of audio, 1 for each microphone so that (I assume) after convert to .h5 the script sees what it expects.
Would anyone be able to point me in the right direction? I'm running this under Ubuntu 22. Thanks in advance for any advice!
Do you mean how did I merge the videos?
I've used the ffmpeg tool.
Others seem to have the same question https://groups.google.com/g/acoular-users/c/D_8rDygNmiI
“we can superimpose the beamforming data over the video” How?
Hi! Could you show the code for the ReSpeaker 4 microphone array (~25 USD) with a GCC-PHAT algorithm? Please answer me,thanks!
Yeah, this would be awesome! @Michael, could you please?
Hi, your work is very interesting, i am also intersted in ReSpeaker and GCC Phat algorithm, can you give me some information? thank You
Which information would you like?
Having a blog post to replicate your result with the ReSpeaker 4 mic would be SUPER helpful. I already have the ReSpeaker 4 mic but found existing tutorials / code difficult to locate sound source.
I want all the information and indications about ReSpeaker 4 to replicate your experiment, too. thanks
possibly all the information and indications to replicate your experiment, thank You very much
Hello, I'm also working on an Acoular project and am trying to use audio samples that have worked to debug my code. Would really appreciate the audio files if possible!
Would be interesting to see what it shows when pointed at a stereo sound source. I.e. whether it will show that the sound is coming from two different speakers, or if it shows that the sound is coming from the “phantom” centre
I would really like to read another blog post about the ReSpeaker 4. While looking for the shield I also found the ReSpeaker 6-Mic. Did you try your Code with the ReSpeaker 4 and a custom MicGeom file as well? How big are the differences?
Are you going to fix your article to show how you superimposed the beamforming image over the video?
You create an H5 format file. Then, you merge a wav and avi file. At what step did you, and how, place the overlay on the video? The article seems to have a large gap at the end.
Hi Michael, this is a fantastic project. I'm wondering if you found the 4 mic array less sensitive to low frequency noise. I also wonder if this is possible with omnidirectional mics if their geometric spacing is known.
Hi Mike, Do you have git repository for the ReSpeaker and GCC Phat algorithm experiment? Would be great of you put it up and possibly others to contribute if they try it and maybe refine and add to it. Just a thought as I am interested to try it and contribute. Thanks Hoo
hello，Can you give me the full code for the UMA-16 DIY Acoustic Camera，thanks
Can the current software capture specific frequencies? From reading your post, it seems the detection is based on the highest amplitude of the pressure wave. What if one is to introduce a tunable band-pass filter to help focus on specific frequencies only?
Is there a possibility of filtering frequencies ? I want to see if specific frequencies can be isolated form same frame.
m01-31-2022 Hi, there. Would you show the code for the ReSpeaker 4 microphone array (~25 USD) with a GCC-PHAT algorithm?
This is very interested! I am intending to do the same thing but in 3D (hemispherical shape). These are some of the hardware so far, please allow me to share
https://photos.app.goo.gl/5zZwiGGPTdpQ1mF19
https://photos.app.goo.gl/GaWTS7Y5j7uqGkG16
Using Acoular, can we make it real time?
Thanks!
When I run the record.sh script and it creates the initial audio.wav file. If I look at it using ffprobe it reports that it has captured 128 channels:
Input #0, wav, from 'audio.wav':
Duration: 00:00:19.37, bitrate: 32768 kb/s
Stream #0:0: Audio: pcm_s16le ([1][0][0][0] / 0x0001), 16000 Hz, 128 channels, s16, 32768 kb/s
[STREAM]
When I convert that to .h5 I assume it retains the same audio channels which results in an error when running the script to analyze the results.
---> 14 pm = bb.synthetic( 8000, 3 )
ValueError: 128 channels do not fit 16 mics
I'm sure this is something really simple to fix, but I'm not familiar enough to know how to alter that initial recording to only capture 16 channels of audio, 1 for each microphone so that (I assume) after convert to .h5 the script sees what it expects.
Would anyone be able to point me in the right direction? I'm running this under Ubuntu 22. Thanks in advance for any advice!
Brock