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

143
hpr_transcripts/hpr3474.txt Normal file
View File

@@ -0,0 +1,143 @@
Episode: 3474
Title: HPR3474: H P R and Audio Fun
Source: https://hub.hackerpublicradio.org/ccdn.php?filename=/eps/hpr3474/hpr3474.mp3
Transcribed: 2025-10-25 00:06:10
---
This is Hacker Public Radio Episode 3474 for 1st April 25th of November 2021, today's show
is entitled HDR and audio fun, it is the 80th show of operator and is about 11 minutes
long and carries an explicit flag for summaries, comments on show and audio processing.
Hello everyone and welcome to the episode of Hacker Public Radio with your host operator.
You know, we talked about audio and kind of response to the community news around intro,
outro, music, whatever.
So I have some comments and replies, the RSS-Future.phb is awesome, I'm going to use that
one instead so that I can get a front of any requests for episodes, I don't follow the
mailing list very well, so by the time I follow the mailing list and see anything in there
then I won't see the call for shows.
So anyways, the RSS-Future is really cool, I did donate to the honest toast donation link
on the about page, I haven't seen that with the HVR forever and I didn't know that you
could actually donate to them, I didn't really want to server but I also wanted to get
to review it.
That's cool, I would say for the donation thing for honest toast, maybe make a different
higher value or a static user input field for that so you could put maximum of $2,000
or whatever, but I would say make it more than 20 or whatever, but I did do the 20, but
I would have done like 50 or something.
Anyways, that's that, the HVR public code on a source get lab doesn't appear to be up
as of today, so what I'd like to do is have somebody reach out to me, especially
if there's existing code out there, definitely can take a look at it, if you guys are willing
to share it or send me a link, I'll double check the mailing list and it's for URL.
But anyways, if there's existing code, I can match that to make runs of automation because
that's what I do.
I automate everything.
Let me know if I can help in that way and I've asked before, but I haven't done anything
active except the post on Bosch and my Texas Thoughts around tagging and all that.
So look forward to hearing some comments on that one around my little script that would
convert audio to text keywords based on anyways, that's that the second thing would be
running a compressor or a limiter, now you can get a hardware based compressor limiter
relatively cheaper cheap that takes like HDMI input even I do it through software, so
I have a media center at home that I use for Cody and that machine will run a there's
pulse effects and there's also which is a flat pack and then there's also a one for
digital audio as a device, as a new device called dynamic range impression, LADS PA plugin.
That's kind of for dynamic range compression, be that or whatever, and I will copy and paste
my notes and all my Cody stuff in the show notes, but anyways, that is available to you.
You can adjust that in Linux for Android, there's a program called, that I successfully
use called Glyclox, G-L-Y-C-O-X for Android that you can have an audio compressor.
I think it's, it binds, it bounds the whole system, so you don't have to like play your
music through that player, prior versions of Android, you had to play the music through
the equalizer to gain control over audio and they've opened some of that up to allow
you to control audio devices with an app, so that's cool.
So it's G-L-Y-C-O-X, that is a compressor which I use for the lows on podcasting, so when
I'm listening to a podcast, I can trim and use that compressor feature and limiter feature
to compress and limit the lows that come through and also kind of not make my ear bleed
when I've got crappy USB sound, and I can make that, I can adjust and EQ that sound to
my liking in real time, so people, folks that have problems with, you know, limiting or
it's too loud, once, once audio, once you have a signal that's reasonable, that doesn't
have a lot of disturbing noises in it, it's a clean sample, and that's sometimes the
problem, but most of the time it's a noise background noise, which can be fixed, or it's
a click pop thing, or like me, I'm moving around and I have a headset on my face, and
that's the most annoying one, it's not constant, it's variable nature, and there is kind of,
it's kind of a low, this is kind of a low quality signal, or a low volume signal, and I'm
having to boost it in real silence, so with that said, if you have it loud enough for
your ears to hear the, of course, voice of a podcast, then once you have it loud enough,
then you can use limiters to equalize it, compress it, make sure that there's no, you
know, loud, you know, disturbing noises in it, and you can save that profile, I'm probably
sure that Andrew, or Apple has the same thing, but I'm assuming most of the folks listening
aren't Apple people, but there's probably an Apple equalizer, not an Apple equalizer,
but you want an Apple compressor or a limiter and, or dynamic audio, we're not talking
about equalizers, we're not talking about, you know, noise reduction, we're not talking
about noise removal, or we're talking about compressing the audio, to where you have
a bunch of layers of sound coming in, and they reach a certain peak, and you can trim
those down, or like, for example, dark mode, or night mode for your receivers, which they
have sensor moved, where you have a really loud movie with really loud music, you can
compress that audio, for example, the blade runner, new blade runner movie has very
loud music in it, and towards the end, it even louder somehow, so that's a really good
example that I use the last scene of the blade runner, one of the last scenes of the blade
runner music with all the water, I will use that to test my compressor, and once they
pass that test, then I will continue listening to whatever it is I was listening to, and
hopes that my ears don't, don't blend, even sometimes not enough, so audio compression
is key, once you have a loud enough signal that you can hear, then you can manipulate
that signal in most cases, and remove noise, if there's silence, or noise gate, or
a noise foe profile, and then if there is a click pop, you can generally remove that
yourself, of course that requires extra work, now what I will say, as far as audacity goes
in any other program, the process of making voice audio clean, probably be simplified
and automated, guarantee you, it can be automated to the point that it will be pretty
rare in the instance where you have that audio, unless the signal, unless the audio can't
be heard in a clean manner, or you have random things like me moving around and making
the poppy noises, those can't be removed automatically, but a consistent electronic and
or noise background, obviously, so what I would say is that we could implement when someone
uploads an episode, could say please allow for at least 10 seconds of silence in the beginning
of your audio recording for automatic noise removal, now you can choose not to do that,
if you know what you're doing, or whatever, you can choose not to do that, and the program
will automatically detect that there is no silence in the beginning of the episode, and
there is, you know, within the first, you know, 30 seconds of the audio, we can say okay,
if within 30 seconds there is no distinct change in the decibel meter, the loudness of
the text, we can say okay, they didn't do the noise removal, they didn't want the noise
removal within that recording, so we'll automatically just trim that down or do whatever we need
to do to post process, but if we want to do automatic noise removal, we can say please
allow at least 10 seconds of silence for automatic noise removal, that will reduce a fair
amount of stuff that comes through this pretty bad, because we can automate that noise reduction
and kind of clean up the audio, and there's a million ways that people prefer the audio
to come in, but I prefer a little bit of bass for low voices, but I don't want my ears
leading when I'm in the car driving down the street and having it loud to the point where
I have to turn it up so loud that the lows on the voice vibrate the whole car, so anyways,
10 seconds of required or suggested silence, I can write a script to detect if that silence
exists or not, if it does exist, use that for the noise profile, automatically span that
across the whole sample, and then we'll have noise reduction, automated noise reduction
within HPR, now the second piece is like just bad, not quality or not being able to separate
that audio out, there's not really much you can do, if it's really bad, bad audio, there's
not a whole lot you can do, if it's mine is probably the peak of this audio, it's probably
the peak of annoyance or quality, about as low as you can get before things, get kind
of irreversible type of deal, so what I'll say about quality increasing the quality, not
a whole lot you can do outside of that, please allow 10 seconds of silence for your recording
so that we can automatically remove noise, and I can do all that, no problem, the second
piece is the intro, the volume of the music and the volume of maybe the text to speech,
maybe we want to glue this together, what it felt like to me is that the audio was a little
high on when we mixed in the audio with the text to speech, so general audio needs to
be background, music or background music in the audio, so especially if they're talking
going on, so you want to have that music be gated out, and then if they're talking you
want the talking to be heard, and then just barely hear that music in the background,
to the point where they can hear it, but they don't necessarily can determine exactly
what the music is or whatever, so you want that music to be as low as it can without
actually hearing it, that's how I feel comfortable, and if I've heard a podcast or any other show,
if they start talking over the music, it's generally disturbing every time, but in where I've heard
it work in some cases, as that music is so low you can barely hear it, and it just provides kind
of an ambiance to the text to speech or the person that's talking, I used to do it, I don't
prefer it anymore, but sometimes it helps with boring voices like myself, the other thing I'll say
I'll give you guys your time back here, I think that's pretty much it, I would say Vosk could help
in that instance, so if a user wants to just have audio, or they just want the voice, we can text
the speech, or speech text their episode, pop that out into a text to speech only, and then they
can use any e-reader or whatever they want to speech to text, the transcribed stuff through Vosk,
so that's pretty simple, we can do that with little to no work based on the episode I did with Vosk
and the automation around, converting media to text and keywords, so I can definitely do that,
but give me a handle on the code, if not I can write code to do that, take some example,
find some example audio from HBR episodes, that's not ideal, and I can try to run some automation
around those, and then just hand you the script and say here, just run this, and it will fix all your
audio or whatever, about the compressor, I think that's pretty much it, I'll let you guys go,
and I might do another one, and we'll move on there, thank you.
You've been listening to Hecker Public Radio at HeckerPublicRadio.org. Today's show was
contributed by an HBR listener like yourself, if you ever thought of recording a podcast,
then click on our contribute link to find out how easy it really is, hosting for HBR,
this kindly provided by an honesthost.com, the internet archive and our sync.net,
unless otherwise stated, today's show is released under Creative Commons,
Attribution, Share Like, Speed of O-Licence.