Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: Kitten TTS – 25MB CPU-Only, Open-Source TTS Model

github.com

Post reply on HN
Print view View original

181–190 of 383 posts

Re: Show HN: Kitten TTS – 25MB CPU-Only, Open-Source TTS Model

#181

Post by devnen »

klipklop wrote: post #91

I tried it. Not bad for the size (of the model) and speed. Once you install all the massive number of libraries and things needed we are a far cry away from 25MB though. Cool project nonetheless.

That's a great point about the dependencies.

To make the setup easier and add a few features people are asking for here (like GPU support and long text handling), I built a self-hosted server for this model: https://github.com/devnen/Kitten-TTS-Server

The goal was a setup that "just works" using a standard Python virtual environment to avoid dependency conflicts.

The setup is just the standard git clone, pip install in a venv, and python server.py.

Re: Show HN: Kitten TTS – 25MB CPU-Only, Open-Source TTS Model

#184

Post by msgodel »

nine_k wrote: post #11
blopker wrote: post #8

Web version: https://clowerweb.github.io/kitten-tts-web-demo/ It sounds ok, but impressive for the size.

Does anybody find it funny that sci-fi movies have to heavily distort "robot voices" to make them sound "convincingly robotic"? A robotic, explicitly non-natural voice would be perfectly acceptable, and even desirable, in many situations. I don't expect a smart toaster to talk like a BBC host; it'd be enough is the speech if easy to recognize.

I personally prefer the older synthetic voices for TTS when the text is coming from software or a language model.

Re: Show HN: Kitten TTS – 25MB CPU-Only, Open-Source TTS Model

#186

Post by dzogchen »

antisol wrote: post #103

System Requirements Works literally everywhere Haha, on one of my machines my python version is too old, and the package/dependencies don't want to install. On another machie the python version is too new, and the package/dependencies don't want to install.

Such an ignorant thing to say for something that requires 25MB RAM.

Re: Show HN: Kitten TTS – 25MB CPU-Only, Open-Source TTS Model

#187

Post by baobun »

hahn-kev wrote: post #126
antisol wrote: post #103

System Requirements Works literally everywhere Haha, on one of my machines my python version is too old, and the package/dependencies don't want to install. On another machie the python version is too new, and the package/dependencies don't want to install.

Python man

    man python
There you go.

Re: Show HN: Kitten TTS – 25MB CPU-Only, Open-Source TTS Model

#188

Post by bashkiddie »

TL;DR: If you are interested in TTS, you should explore alternatives

I tried to use it...

Its python venv has grown to 6 GBytes in size. The demo sentence

> "This high quality TTS model works without a GPU"

works, it takes 3s to render the audio. Audio sounds like a voice in a tin can.

I tried to have a news article read aloud and failed with

> [E:onnxruntime:, sequential_executor.cc:572 ExecuteKernel] Non-zero status code returned while running Expand node. Name:'/bert/Expand' > Status Message: invalid expand shape

If you are interested in TTS, you should explore alternatives

Re: Show HN: Kitten TTS – 25MB CPU-Only, Open-Source TTS Model

#189

Post by 3rd3 »

dr_kiszonka wrote: post #150

Microsoft's and some of Google's TTS models make the simplest mistakes. For instance, they sometimes read "i.e." as "for example." This is a problem if you have low vision and use TTS for, say, proofreading your emails. Why does it happen? I'm genuinely curious.

You probably mean "e.g." as "for example", not "i.e."?

This might be on purpose and part of the training data because "for example" just sounds much better than "e.g.". Presumably for most purposes, linguistic naturalness is more important than fidelity.

Re: Show HN: Kitten TTS – 25MB CPU-Only, Open-Source TTS Model

#190

Post by theshrike79 »

nine_k wrote: post #3

I hope this is the future. Offline, small ML models, running inference on ubiquitous, inexpensive hardware. Models that are easy to integrate into other things, into devices and apps, and even to drive from other models maybe.

This is what Apple is envisioning with their SLMs, like having a model specifically for managing calendar events. It doesn't need to have the full knowledge of all humanity in it - just what it needs to manage the calendar.
Post reply on HN