Other AI enhancements or starts in TANP 6 include pidscan using TSDuck. While adding a spectrum analyzer seems like a grand idea, in reality it is not going to do a whole lot due to the limits of what is available in receiver resources. A proper pidscan though can show things at times that are normally not seen.
Some attempts at a spectrum analyzer date back to the days of Relook 400 or Cuberevo 900 with enigma1 (unfortunately the link to the source code is no longer working). https://legitfta.com/forum/attachmen...7&d=1749284065
Pidscan works well with dvbsnoop so far, although it no longer names some new stream types correctly (for example pid 4094 for DCII).
The first (year 2018) to try using TSDuck in OE was athoik from the image team SDG (SDG Image 6).
It is good to continue this effort.
PS: It wasn't open linux, but once upon a time Dr.HD implemented an analyzer into its regular receiver.
I appreciate seeing that spectrum, but IMO, it would take resources to run it, and it really would not give us much in a FTA receiver unless maybe you had a TBS tuner connected, which would be interesting. TSDuck has been dropped I think, so reviving it again pays homage to Athoik. Work will continue with MODCOD as time permits.
@Arlo,
I appreciate your candid response, and I understand the skepticism about AI - it's completely understandable given how it's often misrepresented or over-hyped. Let me clarify what AI integration in TNAP actually means, because it's nothing like the "I'll be back" scenario you mentioned, and it's certainly not the intrusive "AI search results" that Yahoo and others force on users.
When I mention AI assistance in TNAP development, I'm talking about a collaborative tool - think of it as an extremely knowledgeable programming assistant that never gets tired, never forgets syntax, and can analyze code patterns across thousands of files instantly. Here's what it actually does:
Real AI Examples from TNAP Development:
When working on the wheel loader for Edision driver selection, AI helped identify optimization patterns that reduced load time by analyzing similar implementations across different platforms
For the low symbol rate satellite transponder reception (below 1000), AI assisted in refactoring the demodulator code to handle edge cases that were causing reception issues
In implementing TSDuck integration, AI helped parse through the extensive TSDuck documentation and create efficient bindings for the Enigma2 environment
For blindscan improvements, AI analyzed existing scan algorithms and suggested mathematical optimizations that improved scan speed without sacrificing accuracy
This isn't some autonomous system making decisions - it's me asking specific technical questions and getting detailed answers with code examples. Think of it like having a senior developer who's read every programming book, analyzed every open-source satellite receiver project, and can instantly recall any of it when needed.
What AI Doesn't Do:
It doesn't "take over" anything
It doesn't make autonomous decisions about your receiver
It doesn't collect data or "phone home"
It doesn't change how TNAP operates for end users
The AI integration I'm talking about is purely on the development side. When you use TNAP, you're not interacting with AI at all - you're using code that was improved with AI assistance during development. It's like saying a car was "designed with CAD software" - the CAD doesn't ride along with you; it just helped make a better design.
Regarding the "closing TNAP" consideration - I hear your concerns, and you make valid points. The SDR# example is a perfect cautionary tale. The frustration comes from spending hundreds of hours improving something, only to have people complain about free work they didn't contribute to. But you're right - the community benefits from open development, and spite-closing would hurt the very people who appreciate the work.
As for the manifest and terminal usage - fair point about beginners. Perhaps a simple GUI option to view installed packages would be more user-friendly than expecting terminal usage. The goal is to keep things clean, not to make them unnecessarily complex.
The AI assistance has genuinely accelerated development and improved code quality. It's not marketing hype or forced integration - it's a practical tool that helps solve real technical challenges faster and more efficiently than working alone. No Skynet, no HAL 9000, just better code through enhanced development tools. Hopefully This makes clearer the AI integration into TNAP.
THAT was a very thought out and intuitive response El Bandido! Instead of getting peeled out you made some very clear reasons and explanations. My AI impression may be one sided but I do understand "programs that write programs" all the while debugging and identifying bottlenecks and providing alternates.
On another site there is a discussion as to a receiver that decodes 4:2:2 (it never ends, don't it). It carries on to narrow down to an Apple TV and Nvidia Shield. Specs and capabilities are posted for both.
Although I hate them with a passion. It looks like Apple comes out tops. A fellow there keeps posting a Chat GPT (he calls it "his friend") of the 2019 Shield. BTW I really like the latest Fire TV Cube.
I'm not up on the Shield and a quick specs search comes up with basically what you see on most FTA satellite related vendors sites. Stock release factory info.
Keeping with the subject. It takes folks like you that go in-depth to try and beat the crap out of a devices capabilities and prove true or BS of the printed nomenclature. And you've done one heck of a good job at that.
AI can be approached in many ways now that we are in the middle of the 2020's. Trust your gut. Question "authority". We were joking about knowing where you are driving to and getting a constant.....
REE-Calculating.
Linux applications. GNU. Stuff. Hard work and protecting intellectual properties is fine. True open source projects that get pulled aside by a group with a ton of work given to them. And eventually deemed closed source. All the while "forks" of the same root project which remain open source. And when someone finds a way to write a bit of code that duplicates the functions of the closed source module(s).
With the guys yelliing that the stole it. Did they really? Not so sure.
No need to get into depth. If TNAP is modeled after OpenPlli. Using plugins created by people who do it because they enjoy it. And proud to distribute it. The blindscan plugin. The elusive Services feature in Signal finder. Cool and useful things. I know so little of what it takes to integrate workable variants of any of them for different E2 (or any program, architecture out there) images.
As far as the spectrogram that Eno showed. Choppy, ancient. Still an awesome idea. It would, exactly like you, we, us, would see using an SDR, spectrum analyzer tapped into a LNBF.
The idea of using a TBS card to scan. Looking for wild feeds. Being able to see when a transponder gets turned on and configured from the ground station for an event.
Or just letting a person know a signal is present, modulated or just a carrier wave.
During ball season a few satellites get almost weekly uplinks with freqs. and such all over the place. But it takes work, intuition, patience, and time on an E2 receiver "hammering" the satellite with blindscans finding them. Most don't have a SA to hang off of a tapped coax line. But it would be a rush to see a new spike on the display. Plug the frequency into the receiver. And see what materializes.
The same I would do using a DVB PCIe card in a pc and stabbing in into the receiver, and possibly (me, definitely!) adding it into the satellites.xml file.
Oh. Mista' Kotta', Mista' Kotta' !!! A few posts back. E Channelizer. I'm using an older version. Can select lamedb v4 & v5. I just use the default v4 and always have.
Awhile back I put up a sample of a chunk of the .xml. You responded that if anyone used it, it would corrupt their .xml. Which was 100% true if c/p into theirs exactly as shown.
The E Channelizer channel editor writes satellites.xml with each line starting with 4 padded spaces. A stock, out of box .xml starts with two tabs.
Reveal all (unprintable) characters in notepad++ or whatever XML editor du-jour. And there it is. It still works an any image that I ever used without any errors. You know, to streamline setup. Writing the channel list into a receiver. Init 4 and transferring the .xml into the receiver. Then either a reboot or Init 3 command. So that might just be my issue in TNAP v6 with the alpha ordering. Which BTW after looking into it a bit more. Some channels are in the editor order and some persist in alpha order. So that is my screw-up and up to be to decipher.
Okay. Thanks for the cool AI explanation and lesson. Let me let us get back to the TNAP 6 discussion. Oh. Look out. That Eno is a smart cookie. He's helped me out several times and always very patient and informative. Haha! ahl be bock.
FWIW,
You are not gonna get very far with coding aspirations using a free version of chatgpt or any AI Free Version. Those versions simply exists as demos. The good AI's that can code cost money, and depending on the project, you may need more than one. Coding AI's are subject to errors just like humans, so uttermost attention at all times is needed. But the results, when you get them are nothing short of impressive. There is an art to using AI's in coding. It is not Plug N Play.
So to be clear, we are not using the cracker jack box type Free Version of ChatGPT, we are using coder versions of AI that require money to run. Will get started back on finishing this TNAP6 image as time permits. Have not been able to do anything with it for over a month now, which is unusual.
Originally Posted by el bandido
Reply With Quote
