> in which the schizoid individual can express a great deal of feeling and make what appear to be impressive social contacts yet, in reality, gives nothing and loses nothing ... secret schizoids ... enjoy public speaking engagements but experience great difficulty during the breaks when audience members would attempt to engage them emotionally.
It makes me think psychiatrists have no clue what that schizoid type really is. Rather, they carefully document various semi-related mental qualities, put them in a box, label it DSM-42, and pretend they are doing science.
FYI, it supports decoding ultrasound wav files. I use a Dodotronic UM250K mic to capture ultrasound up to 125 kHz. LCD monitor and other electronics emits some interesting sounds. For some reason, getUserMedia doesn't allow high sample rates, but ffmpeg works:
I've been looking for something like this to review books. Two suggestions: 1. allow to load a book from a url, so notes could be added to arbitrary books; 2. allow to select text and add notes to that selection.
I do have a WebGL-based implementation of FFT, but here I used good old JS. When properly written, it gets translated into really fast machine code, which is even faster than WebAssembly (I tried!). WebGL's problem is the high toll on the CPU--GPU bridge. When you need to transfer a block of audio data from CPU to GPU to perform calculations, you wait. When you need to transfer the FFT data back, you wait. These waits quickly outweight everything else. However on wavelet transforms GPU comes first because you can store some pre-computed FFTs on GPU and reuse them in multiple runs.
Author here. This is a basic spectrogram visualizer that's mobile friendly. It allows to select regions on the spectrogram and play them separately. There is no grand plan behind this web app: it's just a handy basic tool to capture sounds on your phone and see what they look like.
Your spectrogram looks elongated horizontally because the FFT window size is too large. I use window size 1024 with sample rate 48000 Hz, so one window covers 1024/48000=0.02 sec. This window size looks optimal in most cases: if you change it in my web app, you'll see that all other window sizes get the spectrogram blurry in different ways, but at 1024 it gets into focus.
Of course, don't forget the window function (Hann, or raised cosine), but it looks like you've got that covered because your spectrogram looks smooth.
The color palette looks good in your case. FWIW, my color function is like this: pow(fft_amp, 1.5) * rgb(9, 3, 1). The pow() part brightens the low/quiet amplitudes, and the (9,3,1) multiplier displays 10x wider amp range by mapping it to a visually long black->orange->yellow->white range of colors. Note, that I don't do log10 mapping of the amplitudes.
It uses "audio/webm;codecs=opus" to record mic. Now it's possible to change it in the config menu in the top right. Safari probably needs audio/mp3. Edit: also consider "audio/foo;codecs=pcm" where "foo" is something compatible with Safari.
Very neat! May I suggest adding a button to switch to log scale for frequency? I love the ability to select and play back just a particular set of frequencies. But voice uses only about ~15% of the screen height [1], so it's hard to play with.
Zooming is not really a way to get what I'm after, because I was trying to hear particular bands one after the other. E.g., trying to listen to one octave after the next. And since the octave relationship isn't linear, I'm thinking a non-linear scale would better match what I was trying to do.
The WebAudio API has an anlayser node that can create spectrogramms in real-time. The ones I've created in the past were nowhere near as detailed as this one though.
Can I ask what kind of use cases would a spectrogram have for radar data? I've been messing around with making my own spectrogram app as well (linux desktop app and not web app though) and would be stoked to know if there's any potentially easy to reach use cases for it
We basically make doppler radars - here frequency-shift is proportional to the speed of the object. Most other radars (pulse-radars) uses the bandwidth of a pulse to gain range resolution (the wider the bandwidth the better the res).
Radar signals are modulated in very specific ways, which are visible right away even with plenty of noise on a spectrogram. Classifying the modulation of radar signals is something common in military contexts, since it allows you to listen for emissions and be able to tell if it's an enemy or an ally. I bet it has more uses than that, but it's the first one I could think of.
Regardless of the current SOTA, this is a task where I'm fairly confident future LLMs can get very good at, unlike the "code in natural language" nonsense.
I'm thinking something like a linter with access to the AST, that can produce warnings like "you forgot this corner case", "you are not freeing resources. Is this intended?" and so on.
Probabilistic linters does seem like a fruitful realm for them, yeah. A lot of code shares strong similarities with others in small scales, thanks to stack overflow and various other "help me solve X" -> "try Y" answer pairs.
I do wonder how to reliably tell it to ignore noisy, incorrect warnings though. They're potentially sensitive to any new input / weight / random-seed changes, so it seems like literally every LLM upgrade will run the risk of ignoring existing suppressions (or you say "ignore this whole line" and miss useful warnings) due to small perturbations...
