Any alternatives to notebookLM's podcast creator?
Posted by Charuru@reddit | LocalLLaMA | View on Reddit | 24 comments
Great audio output that doesn't sound robotic.
Google's product is pretty good, just the censorship and political correctness is killing me.
Having it discuss a book, whenever a female character does something it goes on and on about how it's so great that it's a female character with agency (honestly feels misogynistic as it poses having no agency as the default).
Can suno or something do this?
vmehmeri@reddit
Check out https://podfeed.ai, it's focused on the audio generation part but with more customization.
Here's one sample: https://storage.googleapis.com/podfeed-public-bkt/audio/seoul-declaration-dialogue-en.mp3
(There are a few more on the app's landing page)
Disclaimer: it's my startup. Lest anyone say I just copied NotebookLM, I started this before NotebookLM launched. The irony is that I used to work for Google until very recently and was completely blindsided by NotebookLM. But well.
Anyways, I will be the first to say that the audio quality doesn't sound as natural as NotebookLM's, but here's where I think I've done something better:
1 - More audio language options (40+ languages)
2 - More user controls: you can specify between monologue (single voice) and dialogue, length, level (NotebookLM tends to dumb everything down which is what annoys me the most), and emphasis (what to focus on). We'll be soon adding the ability to choose specific voices.
3 - The UX is entirely focused on audio generation, as opposed to being a note-taking app that does a lot of things including generating audio.
4 - We have one nice feature which is that you can add the audio to a private podcast feed we automatically generate for you, which you can add to your favorite Podcast player. We're also working on generating a shareable link.
5 - As it's typical of Google consumer products, they're probably collecting a lot of data that you put in. Google has very strong security and privacy practices, of course. But remember, if something is free, you're the product =)
Also, for what it's worth, my vision for this product is for it to help you transform any kind of content into a format that works best for you to consume. Audio is what we're starting with, but we have some ideas we're cooking. I'm also in talks with Google so that we can be one the first ones to try their newer voice models when they release them as an API commercially, and we then we'll add that to the product.
We're unfortunately unable to offer this for free (we're not Google), but we're offering a 50% discount on the first month, and you can cancel anytime. Give it a go, tell me what you think. If you think it was a complete waste of your money, reach out via the contact form and ask for a refund within the first 7 days. Just please tell us why.
Admirable_Common_511@reddit
Just a heads up your Contact Us Form is broken
Crisper026@reddit
I tried to sign up but get an error message whenever I try and select a plan
vmehmeri@reddit
Hey, sorry about that. We had a problem with Stripe that has now been fixed!
5555dimitri@reddit
is there a trial version / free generation tokens? just to see if it suits my needs. I've checked the website and it looks very promising, especially the ability to adjust the learning level! but then I see instant paywall and feel a bit apprehensive to proceed ...
vmehmeri@reddit
hey, totally understand the apprehension. We're not able to offer free trials at the moment unfortunately (we're self-funded), but we're giving 50% off on the first month. If you go for the Basic plan, that means $5 for the month. Would that be reasonable for you to try?
We're in the middle of revamping the UX also so you will be able to access the generated shownotes also for the audio on the web app (which includes key takeaways, suggested action items, and relevant links found within the content - something already available when you add an audio to your private podcast feed, but not yet on the web), share the audio with others, and other nice new capabilities we're cooking. All coming soon.
MustBeSomethingThere@reddit
I am creating a fully local poor man's version of podcast creator.
Here is my first attempt podcast: https://vocaroo.com/1gtEhJkNXStf
Gloomy_Presence_9308@reddit
Yo, don't give up! Keep us updated!
MustBeSomethingThere@reddit
https://github.com/PasiKoodaa/AI-PodcastifyAI
I did that, but I have a lot of other things to do, so I haven’t had the time or interest to continue developing it.
nonerequired_@reddit
Cool
joaobrasil65@reddit
Good but not nearly as good as Google
Wide-Neighborhood229@reddit
So guys, I'm using the LM notebook. However, I'm having great results. I'm leaving the link to my podcast, it's on Spotify
https://open.spotify.com/episode/42soI7EqR3hzFRRQSq9gll?si=stZfbwCURd-hN4VG7HrVEA
It's actually a hybrid podcast, where I'm the interviewer, chatting with the audio generated by the LM notebook.
Candid-Swing-4825@reddit
You can have a look at: Learn with Podcast - Generate Engaging AI-Powered Podcasts | StudentAITools
LuaparK@reddit
Had exactly the same question, you found anything?
cryptopelado@reddit
OP have you found a better solution since you posted this?
vmehmeri@reddit
Check out https://podfeed.ai, it's focused on the audio generation part but with more customization.
Here's one sample: https://storage.googleapis.com/podfeed-public-bkt/audio/seoul-declaration-dialogue-en.mp3
(There are a few more on the app's landing page)
Disclaimer: This is my startup. See my comment on this post
No-Statement-0001@reddit
Here's a demo I created using elevenlabs: https://vocaroo.com/18LwtDCmC4nh
Compared to NotebookLM, there's a lot to improve. Besides the voice quality there is also the script and the interactions between the hosts. The subtle laughs, acknowledgements, follow up questions, topics chosen, all of it comes together to make a very impressive experience.
cryptopelado@reddit
Any chance you share how you did this? Assume I can create the dialog - how would you leverage elevenlabs to create the conversation between both, and then produce a single audio file? Trying to create some dummy podcasts and trying out what works better than Notebook LM
No-Statement-0001@reddit
I used ChatGPT produce a transcript for the two hosts with structured JSON output. Then I wrote a small python script that runs each host's dialog through the elevenlabs API. Finally used ffmpeg to stitch together all the MP3s into a single file.
cryptopelado@reddit
Nice. I've already also built a solution for the transcript, but missing the elevenlabs and ffmpeg part. Not sure how good it would be for podcasts tbh, missing a bit more of the conversation piece of the dialog. If you can share your code in Github or somewhere I would really appreciate that
Charuru@reddit (OP)
Right I don’t think 11labs is it yet.
chibop1@reddit
Use any LLM to output text, and feed that to elevenlabs.
kriskoeh@reddit
This is nothing even comparable.
rbgo404@reddit
We have created Speech to Speech voice bot using all OS models.
Switch to any LLM of your choice!
Here's the blog: https://docs.inferless.com/cookbook/serverless-customer-service-bot