Someone please help me build this:

**Text-to-Speech reader app for Android**

**Instant Playback**: Integrates seamlessly into Android’s share menu, enabling one-tap access to convert text from any source (web, PDFs, EPUBs) into spoken audio. Opens shared text immediately, using the device’s **local TTS engine** (no cloud dependency, using Sherpa TTS) to read content aloud while dynamically highlighting words for visual tracking.

**Queue Management**: “*Share to Queue*” option lets users compile a playlist of text snippets, articles, or documents for continuous listening. Prioritize, reorder, or remove items, with playback resuming automatically between entries.

**Podcast-Style Background Export**: Convert text to MP3 in the background, saving files to a user-designated folder for later listening in a podcast app. Ideal for commutes or offline use.

**Lightweight & Private**: No subscriptions, ads, or data mining. Processes all text locally, ensuring speed and privacy.

This is similar to this non-free app: https://hyperionics.com/atVoice/

Reply to this note

Please Login to reply.

Discussion

Yes, what's needed is a companion app that loads the shared text and sends it to Sherpa for audio playback.

Interesting! What would you use that for? Select text on a blog, share it to local queue and listen to it while on the go?

Exactly.

This would be great. Closest thing to this is speechify on iOS which I use a lot, but its very cloud based and paid

I'll build a Nostr based version of this in the early summer, when we get to Audio widgets with :Zaplab: Zaplab.

Letting you share all that audio in niche communities around certain topics.

Plus, letting you pull in YouTube, Pods, etc besides text. On to the groups relay + server.

Probably calling it Zapcast 😜 .

Nice, let's make it happen.

I remember you (and me!) needing all this, since over a year ago.

Finally getting to a point we I can actually see the #Zaplab team build it.

And if nostr:npub1wf4pufsucer5va8g9p0rj5dnhvfeh6d8w0g6eayaep5dhps6rsgs43dgh9 keeps bugging me enough about Audio widgets, we'll get there :PepeLaugh:

It should just be a DVM.

No, it should work 100% offline.

:Zaplab: Zaplab = :110percent: #localfirst

What's the use case.

Read stuff online?

Think about what you just said...

The challenge here is the interface more than where the TTS engine runs, if it can reasonably be done locally then it should

Yeah, for like a page, that makes sense. But for conversion of PDFs of books and documents into an audiobook, a DVM should be used as that is significant compute. Also, if a DVM is used, when the audio is converted once, the compute doesn't have to be redone unless someone wants it in a different dialect. The interface for webpages is also better with a DVM because if you do it as a webextension or something like highlighter I think then if you visit a webpage and it has already been turned into audio, you can then not have to redo the compute. If this is possible to do locally, then whats more important is getting dvms running on phones like citrine is a local relay.

Sherpa runs in real time, so it would be at most a couple hours for a book, should be fine in the background of a powerful phone.

OK. Used Sherpa for the first time today... These are just features you can add to libera FD, right?

After trying to see how it performs on a 1000 page book, I'm back to thinking this would be better as an application on your start 9. Still thinking it's better as a DVM (if you run your own its still private). You could have a chaim of dvms that you run, 1 to convert PDFs and such to eBook or whatever format you want,

1 to convert it to audio in whatever dialect you want. Maybe 1 for translation and another for adding the chapters.

Otherwise you can just build your standard bookmarking or note taking app like Obsidian I think?

Idk, I see the use case, but it's in general is for hands free stuff so it'd be really weird to want to use it on anything shorter than 2 pages?

Not yet.

As for Nostr login - it would be cool, but there's no value there - you can just ask browser to generate a random username and password. There's no e-mail or anything. I don't even need to know your identity.

Then why have a login?

Because the feed is personal to you, you don't want to listen to someone else's articles. You just add it as a private RSS feed to any podcast app and listen to your articles and keep track.

The app generates RSS feed, your podcasting app can be the used to queue, remember position, etc.

Just try it, just make up a username and password and you're good to go.

Yeah, it seems to work pretty good, and has some of the features you're discussing, including the playlist.

Thanks for the recommendation, I didnt know about the app.

I don't like the UX at all though...

Yes I agree the ui isn't tight, but if something has all the features you want, it's easy to fork it and change the ui

Agree that some compute needs to be done server side, the advantage of doing it locally is privacy. Maybe you don't want to advertise which books/articles you're reading, or doing it 100% offline.

And while some DVMs may offer past computations for free I would not take it for granted.

Lol, typo, offline of course.

I have found DVMs to be a nightmare. This may help you though:

https://huggingface.co/spaces/hexgrad/Kokoro-TTS

The TTS engine will be Sherpa, runs well on the phone.