lee/hpr-knowledge-base
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Initial commit: HPR Knowledge Base MCP Server

Browse Source 
- MCP server with stdio transport for local use
- Search episodes, transcripts, hosts, and series
- 4,511 episodes with metadata and transcripts
- Data loader with in-memory JSON storage

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
Lee Hanken
2025-10-26 10:54:13 +00:00
commit 7c8efd2228
4494 changed files with 1705541 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

134
hpr_transcripts/hpr3698.txt Normal file
View File

@@ -0,0 +1,134 @@
Episode: 3698
Title: HPR3698: Spectrogram
Source: https://hub.hackerpublicradio.org/ccdn.php?filename=/eps/hpr3698/hpr3698.mp3
Transcribed: 2025-10-25 04:16:11
---
This is Hacker Public Radio Episode 3698 for Wednesday the 5th of October 2022.
Today's show is entitled Spectrogrum.
It is hosted by Klaatu and is about 16 minutes long.
It carries a clean flag, the summary is, Edit Audio as a Spectrogrum.
Hey everybody, this is Klaatu and I wanted to talk today about why I edit sound as a spectrogram.
So in Audacity, if you've ever used Audacity and if you're listening to the show, you may have also contributed to the show and if you contributed to the show, there's a high likelihood that you've used Audacity.
But in Audacity, there's this really interesting option to view your sound, not as a wave form but as a spectrogram.
And a spectrogram is a psychedelic-looking representation of data.
You might think of it as a heat map and I imagine a heat map is probably a form of spectrogram.
I should probably know that, I should have looked it up.
It all kind of just came to me right now though, but I'm pretty sure that would be true.
The show shows data by concentrating a color in areas where there's a lot of data and then by starving other areas that don't have a lot of data.
And so in other words, you have sort of, if you've ever seen like, well, like a grayscale or I guess even a rainbow, but let's go with a grayscale where it goes from black to sort of 50% gray, white, that spectrum.
One end of it might be dense with data and then the other end would be sparse on data.
And I think typically because the white end of the spectrum is by definition a lot of data.
Like you get white light on a screen by combining all of all the pixels are on.
All the colors are being combined.
That's different than in physical media.
If you mix a bunch of colors in real life, like paint, you usually end up getting something pretty dark.
Like basically a black might not be pure black because there might be lots of different pigments in there.
It depends on what kind of paint you're mixing and so on, but that's different in projected light.
All of the colors on.
So that's all of the data, 255, 255, 255 in RGB.
That produces a white light.
Turn all those pixels off and you get, you get darkness, you get black.
So in a spectrogram, you might say, okay, well, any place that's really white, white hot.
There's like, there's stuff there.
That's really intense data.
And then if it sort of starts to trickle down into maybe a little bit of gray,
well, we could say that there's still a lot of data, but it's starting to become sparser.
Not all of the data is not all the slots for data are turned on.
And then you keep going down the scale, down the scale further and further.
You're essentially removing data intensity from that region until you get into just darkness,
where there is no data.
None of the data switches are turned on there.
So that's a spectrogram.
And you get that option in Audacity.
You can look at your sound as a spectrogram.
And to read the spectrogram, the spectrogram in Audacity,
you're looking at a couple of different things.
You're looking at where the sound is within the EQ, within the frequency of the sound,
where that data is.
Now, you're not getting the wave form of the sound wave,
but you're getting notation through not intense colors to show you where the sound
that you're hearing is being produced, where that's sort of hanging out.
So for instance, if you are seeing a lot of bright yellows and whites down at the bottom,
then you're looking at bass tones.
You're hearing, you're looking at the lower frequency of sound.
And then as you move vertically up from that bass line,
then you start seeing, I don't know, maybe purples and oranges.
I mean, it obviously, I'm describing color as if they're all spectrograms are the same.
You can change the theme of Audacity and get different colors.
So it may not be that for you.
But they're in the middle, in the mid-range, you're starting to see the mid-range of sound.
That's sort of what you're seeing there.
And as you speak and as you hear a voice fluctuating and changing over time and across different words,
you'll see different colors sort of appear there and sort of different intensities appear
at different levels of that vertical scale.
And then finally, near the top of the screen or of the track,
you'll start to see maybe other colors that indicate the higher frequencies.
And I started editing in spectrograms because, really because of that,
because you could really see where the different sound was coming from.
And I found that it was absolutely priceless for removing all of the inhales that I do.
I mean, I inhale now that I've said it, you're not going to hear anything but that.
And I apologize for that.
But when I'm speaking, I inhale a lot, I feel I'm always taking a breath in.
As if though I'm short of breath, constantly short of breath for some reason.
I don't know why I do it.
I never knew I did it until I started editing myself in podcast form than I could hear.
Another thing I do frequently is, you know, I mean, anyone, anytime you're speaking into a microphone,
you can throw all the pop filters and things on there that you want.
But there's still those wonderful, plosive sounds and those little speckles of spittle.
And all the unpleasant things that happen in a mouth, it just comes through in this audio graph.
And, you know, part of cleaning up podcast audio or audio for anything is removing all of that sort of excess sort of too human for comfort kind of sounds.
And for my money, the spectrogram is the way to do that because you can really zero in on exactly what you want.
I mean, I'll admit that with the audio, with the waveform, I was getting really good for a while a long time ago at recognizing my own sort of cadence
and the parts where I would take a deep breath that I didn't want or that I would say, um, that I didn't want, you know, those things.
All the, I started recognizing those and I was probably up to like, I don't know, 75% 80% accuracy.
And I could pinpoint things and say, that's going to be a little bit of a breath. I'm going to just cut that right out done easy.
Now, if I was editing other people, I didn't know, I didn't know sort of their signatures yet. And so it wouldn't be as effective.
With the spectrogram, it is just so much clearer.
Like you can really, really see the inhales. You can see the exhales. You can see the, the, the clicks of the mouth or the tongue.
You know, you can, you can see all of that. They're, they're really, really clear. And you can select them in the spectrogram.
You can zoom in on a spot where, I mean, you do have to, you have to kind of develop a skill to see them, you know.
And sometimes they're quite small. So, so really you have to hear it first and then you zoom in on this space where you heard the, the click or the pop or the, the, the breath or whatever.
Well, the breath actually you don't really, those are really clear, like really, really clear.
So I'll select a breath point and then hit ZZ for zero crossing is what it's called.
I don't know where that is in the menu or if it's a custom thing that I do, it's got to be under select.
Yeah, select at zero crossing. So select menu at zero crossing. So you click, click that or, or if you have a keyboard shortcut assigned to it, you do that.
That just kind of ensures that you're not clipping things. Now, the interesting thing about these spectrograms is that you're not seeing the audio waves.
So when you're cutting things out, you're not, it's not the same as selecting the peaks and the valleys of an audio wave.
You are selecting a region of sound, the same data, but by different measure. And so you can't exactly like the, the data that you're viewing, it isn't the same as it doesn't translate as a, as a waveform.
And you can even see that, like if you selected a region and then you go back, actually that one didn't work for me.
If you go back and view the waveform or if you turn on multi view, where you actually are seeing them both, which I don't, I don't like doing it.
I don't like the multi view that much. I mean, it is, I can see it being useful, but for me, it starts to distract me and get me confused as to what I'm seeing.
I just prefer to look at the spectrogram myself, but you select an area, you do zero crossing to make sure you're not clipping a sound wave off at an inconvenient point.
And then I just go to personally, I just go to affect amplify and knock it down to zero. That's how I, that's how I get these, these elements away.
Or I just delete them, Z and then X to delete. That's what I use to delete X. You could use whatever you want, but I find it easier to just have it all on that one row.
And then, and then you, and then you go on to the next offensive ugly sound that you have made with your lips and, and cut that out.
So the spectrogram, I don't know, I highly recommend at least taking a look at it. If for no other reason, then to experience the visuals of sound by some other representation.
And I didn't really fully appreciate that until I started editing spectrograms. I'd seen the spectrogram before and it just looked different and I figured it was probably some kind of, I don't know, data science, scientific thing of interest or something.
But when I started to actually use it for editing, I realized that I was getting the same, I was looking at the same data as I'd been looking at through waveforms.
But I was getting so much more information somehow, like I was, I was seeing a much more complete picture. Now, obviously that, that's for my purposes.
I imagine spectrograms are not useful for a lot of other activities. And I mean, certainly I don't, I don't know that I could, I mean, I haven't tried.
I don't know that I would try to edit music, like a music track through as a spectrogram. Maybe I would. I don't know. I don't think that would work so much.
I mean, I guess it depends on what kind of, if it's just a bass drum or something, I'm sure you could probably pick those out pretty, pretty easy.
But for, for what I'm doing, spectrogram has been, has been just really, really, really useful. And just a fascinating study at kind of looking at, at different, at different views of, of data that, that I thought I was pretty familiar with.
So yeah, next time you go to edit something in, in audacity, I mean, as long as you're not against a deadline or anything, if you've got time to kill, take a look at the spectrogram, take a listen to it, try selecting some areas, get that zero crossing, silence it or cut it or something, and just see what you've gotten.
You might really surprise yourself. I have been surprised in the past of just how, you know, you can just kind of go in and zero in on something that's, that's in an area that you don't think it should be in, you select it, you delete it, and then you listen to it back thinking, well, I've just taken a big chunk of audio out.
Obviously I'm going to hear, I'm going to hear where I, I, I cut that out, and no, you don't hear it, because you're not, you're not working with the waveform, you're not, you're not just clipping out this, like, a word as you, as you, I think you, we often zero in on the waveforms, you're, you're clipping out frequencies in a way.
I mean, ultimately you're affecting the waveform, I understand that. But the, the selection process is, is different than what you might expect.
And it's kind of interesting, because I, I do, as I often do, I extrapolate a lot of this into, into the, into my real life. And I think about how my perception of audio has changed, and it strikes me that sometimes we are set in a certain mindset.
We, we have a certain interpretation of our surroundings, and it seems very correct to us, and it seems like we have fought this through, we've listened to people that we trust, and, and that we form ideas, we form these interpretations of our reality.
And then you, you, you learn to look at something a little bit differently, and all of a sudden, I mean, you're looking at the same exact data, but you're recognizing it to be something, you're, you're recognizing it as, as taking a new form.
It's, it's literally the same data, but you are seeing aspects of it that you, you just couldn't see before, and that to me is absolutely fascinating. It's obvious, it's something that has not, it's been said before, like, that's not a, a surprising thing.
Oh my gosh, you can get a new perspective on things and change your mind. I'm just saying, I hadn't, before I started switching over to spectrograms, I hadn't really seen, it, have such a pragmatic effect on, on what, how I interact with something.
The way that I interact with sound now is, is different. It's literally different than it had ever, it, it had been before. And, and across my entire life, my understanding of sound has, has actually changed a lot. It's evolved a lot.
Started out, probably as sort of an understanding of, I mean, without going into child psychology, which I know nothing about, I'm sure there were some, at some point you learn about sound, like, I don't know how that happens, and I don't care.
At some point, you know, playing in a high school band or whatever, you, or at least I sort of understood sound as musical notes, like that was kind of like the representation of sound to me, because that's what you did to make sound, you, you read the notes and did the thing that that meant for you to do.
And then I, at some point I learned about guitar tabs, and that completely took me by surprise. I just thought, some people read, you know, read music by these charts.
Like, how is that, how is that possible? And then, you know, at some point I obviously learned about sound waves and, and wave form representations of those, and learned that those could be affected through electronic means and so on, or through other means too, acoustic means, put something in a different room, you, you'll get a different sort of experience of that.
And then eventually discovering spectrograms, which, which completely shifted everything again. And, and that's just, that's just a little tiny snapshot of one of one understanding of, of a piece of data of, you know, of one attribute of reality.
So that to me has been very, very fascinating. And it's, it's interesting to me that even something as simple as just kind of like, yeah, every day experience for, for most people of just how we hear things. And then for a smaller group of people of how we, how we visualize what we're hearing on a computer screen and how we're interacting with that can completely change based on on how you're looking at that data.
The data itself hasn't changed. You are simply looking at it through a different lens. And that, that changes again, not the data, but it changes how you interact with it. It, it changes possibly how quickly you edit or how cleanly you edit or how the things that you're able to zero in on whatever.
Well, that's it for me. Thanks for listening to this episode of Packer Public Radio. You should record your own, go edit it in audacity with spectrogram. Talk about your experience. Talk to you next time.
You have been listening to Packer Public Radio at Packer Public Radio does work. Today's show was contributed by a HBR listener like yourself. If you ever thought of recording podcasts, you click on our contribute link to find out how easy it really is.
Posting for HBR has been kindly provided by an honesthost.com, the internet archive and our syncs.net. On this otherwise stated, today's show is released under Creative Commons Attribution 4.0 International License.