Podcast adblock

[apoyusiken]

Is there a way to block builtin ads on podcasts? Sponsorblock on youtube could work but youtube isnt exactly the best app for podcasts.

[whispous]

Well, you'll need to pick which podcasting app you're looking to use as there won't be a single solution that fits all of them.

 

Additionally, some adverts in podcasts are just directly part of the recorded audio, with no markers that a plugin/extension could latch onto to detect.

[GuiltySpark_]

If the app you use supports extensions and there is something like sponsorblock, but I highly doubt this exists. 

 

I'm going to say no.

[maplepants]

Not directly. But A lot of Podcast players (like Overcast, the one I use) will have a setting per podcast which is "Skip the first x seconds of each episode". Obviously that does nothing for mid episode ads, but it does help with big podcast network stuff where they often start each episode with 1-2 min of ads.

[OddOod]

Fingers crossed adblock discussion is allowed on the forums. That being said here goes! 

Possible? yes. Easy? lol, *far* from it. I am assuming you are consuming podcasts directly from a feed rather than via Youtube. Sponsor block is perfectly fine for YT assuming it's popular enough for someone else to do the labor of tagging ads.

As someone who gets through ~100hours of audio a week, it's something I've put some thought into

First off, you'll need to figure out a way to rehost all your feeds. Not rocket science, I've done it and I'm not that good. 
I built a lil tool to split up the mega feeds of a few podcast networks into constituent shows. It's clunktastic and has to be manually updated when new shows debut, but it works remarkably well. Fundamentally a podcast feed is just an XML file in a specified format which includes at least a title, a publish date, and a link to an audio file. The script I wrote just takes the original feed, filters on title, and spits out the results into new files, all in a folder. Then I use a *very* simple python script to host that directory as a web directory. EZPZ.

So, now you have the feed hosting set up, you need to do the actual ad removal, and how you do that is gonna vary from podcast to podcast, but all of them are going to require you know how to build signal processing pipelines. Not exactly rocket science, but also not throw an afternoon at the project easy. Some podcasts extremely kindly break the episodes into chapters, and some of those chapters are even literally labelled as ads. For those, you just have the pipeline snip out the ad section, stitch it back together, and spit out the file. Other podcasts aren't as nice, but do us the kindness of having an ad transition chime/riff/chord/tone. For those, your pipeline merely has to find the in and out tones, then snip between them. Again, pretty simple, though it will take some CPU power, you won't be running this smoothly on a Pi. 
That takes care of the easy ones, but in my experience they are by far the minority. Most podcasts don't use standardized transitions and further use dynamic ad insertion. The upside of that case is they don't tend to rotate the ad stock super frequently and those ads will all be identical. For those cases, you will have to program it to compare several episodes, find the sections in common, then snip those out. This will likely also snipe intros and credits which may be a boon or a bane depending on your taste and which shows you follow. A further potential issue is that if the pipeline only makes one pass, you will likely hear each ad once, the first time it's aired on one of your feeds. If you set up a periodic archive scan and you don't listen to eps as soon as they drop, you can probably avoid that, at the cost of more CPU cycles which are admittedly cheap. 
Now, that takes care of the vast majority of ad-supported podcasts. But there are still some out there where the host(s) actually read the copy new for each ad. For some, if they're really pro at it, it will be similar enough that the dupe matching might snag them, but you're in dangerous territory; loosen the match% too much and you might lose parts of the episode. At that point, to achieve total ad-free nirvana, you're gonna need to run the episodes through transcription (I have used Whisper to great success), then feed the transcript into an LLM which has been trained on finding ads. Then you get the timecodes and snip at appropriate sections. 

Once you have your de-ad'd episodes, you give them appropriate names, store the file locally in a hosted folder, update the episode audio URL, and boom, total ad-free podcast feeds.

Okay, yes, that's a lot of work, but that's pretty much what you have to do. Honestly, it seems like a fun project, but I am so loaded with projects at the moment that I can't see fitting it in. Also, I'm a big fan of most of the podcasts I follow. I want them to be around next year. So I don't mind paying a bit of dosh to support them directly for an ad free experience. Definitely a route you should consider.

If you do want to dive into this, there are several projects on github which leverage FFMPEG to do the matching part

[OddOod]

8 minutes ago, maplepants said:

Not directly. But A lot of Podcast players (like Overcast, the one I use) will have a setting per podcast which is "Skip the first x seconds of each episode". Obviously that does nothing for mid episode ads, but it does help with big podcast network stuff where they often start each episode with 1-2 min of ads.

Ugh, the WORST is when they *sometimes* have pre-rolls. I also like Podcast Addict's "skip chapters which fit this pattern" feature where I just match "[Aa]d -*"

 

[GuiltySpark_]

4 minutes ago, OddOod said:

--snip--

 

I think i'd live a full and happy life just hitting the skip 30 seconds ahead button a few times here and there than go through all that. LOL

[notabitail]

I usually just skip the ads using the buttons in the podcast player. 

[maplepants]

34 minutes ago, OddOod said:

Ugh, the WORST is when they *sometimes* have pre-rolls. I also like Podcast Addict's "skip chapters which fit this pattern" feature where I just match "[Aa]d -*"

 

Regex based chapter skipping is a very cool idea! I find German podcasts are great about chapters, but I find that on the whole it's hit or miss whether podcasts I like have them. 

 

The automated ad insertion services big podcasts have are so garbage. The ads are usually so much worse than read ads, and I know from articles I've read by people who write podcast apps that MP3s with these dynamically inserted ads typically violate the spec in a number of different ways.

[apoyusiken]

10 minutes ago, notabitail said:

I usually just skip the ads using the buttons in the podcast player. 

yea but sometimes you cant access your phone

[OddOod]

46 minutes ago, GuiltySpark_ said:

I think i'd live a full and happy life just hitting the skip 30 seconds ahead button a few times here and there than go through all that. LOL

And I do. Though, man, The Memory Palace kills me. LOVE the show, but there's about 6 minutes of content in a 15 minute episode with the rest being ads. I give myself a sore ear with how much I have to tap to skip when I binge it. Though I now see that there's some possibility of paying for an ad free stream!

[OddOod]

Just now, apoyusiken said:

yea but sometimes you cant access your phone

That's why we have controls integrated into headphones.