Kitten TTS
AI
text-to-speech model
SHOW HN2025
1,003points
361comments
#15of 3,837 in 2025
Top 1%of 2025 by points
How HN took it
MixedNo clear lean
33 positive, 76 neutral or asking, 8 mixed, 50 critical, out of 167 comments about the project. Thread score -0.04.
Performance: 7 for, 18 againstExecution: 6 for, 18 againstIdea: 10 for, 5 againstDesign: 2 for, 5 againstDocs: 0 for, 3 against
The thread, comment by comment
Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.
positive neutral or mixed critical about another commenter or something else the maker
What HN said
What people liked
Impressive quality for a model this small
From 3 comments
Web version: https://clowerweb.github.io/kitten-tts-web-demo/ It sounds ok, but impressive for the size.
blopker on HNIt is not the best TTS but it is freaking amazing it can be done by such a small model and it is good enough for so many use cases.
victorbjorklund on HNNot bad for the size (with my very limited knowledge of this field) ! In a couple tests, the "Male 2" voice sounds reasonable, but I've found it has problem with some groups of…
tecleandor on HN
Runs on CPU without a GPU, enabling offline embedded use
From 2 comments
The headline feature isn’t the 25 MB footprint alone. It’s that KittenTTS is Apache-2.0. That combo means you can embed a fully offline voice in Pi Zero-class hardware or even…
MutedEstate45 on HNThis feels different. This feels like a genuinely monumental release. Holy cow. Very well done. The quality is excellent and the technical parameters are, simply, unbelievable.…
junon on HN
Objections
Installation and Python dependencies are difficult; package is far larger than 25 MB
From 4 comments
``` 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…
antisol on HNI 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…
klipklop on HNTL;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…
bashkiddie on HN25M ? lol . the venv is 6.9G
Piraty on HN
Voice quality sounds mechanical or tinny compared with larger models
From 3 comments
Hmm the quality is not so impressive. I'm looking for a really naturally sounding model. Not very happy with piper/kokoro, XTTS was a bit complex to set up. For STT whisper is…
wkat4242 on HNCool. While I think this is indeed impressive and has a specific use case (e.g. in the embedded sector), I'm not totally convinced that the quality is good enough to replace…
sandreas on HNThe samples featured elsewhere seem to be from a larger model? After testing this locally, it still sounds quite mechanical, and fails catastrophically for simple phrases with…
ricardobeat on HN
Requests
Support for other languages
From 2 comments
This great for english, but is there something similar for other languages? Could this be trained somehow to support other languages?
spapas82 on HNWould love to se something like this trained for multilingual purposes. It seems kinda like the same tier as piper, but a bit faster.
gunalx on HN
Audio samples on the project page
From 3 comments
Okay, lots of details information and example code, great. But skimming through I didn’t see any audio samples to judge the quality?
onair4you on HNHave you considered adding some 'rendered' examples of what the model sounds like? I'm curious, but right now I don't want to install the package and run some code.
dirkc on HNIt's just so annoying and idiotic that there aren't a few samples on the home page. It didn't occur to you that it's the very first thing people would want to hear?
OrangeMusic on HN
Training code and data release
From 2 comments
Hi. Will the training and fine-tuning code also be released? It would be great if the training data were released too!
maxloh on HNWhere does the training data come for the models? Is there an openly available dataset the people use?
pkaye on HN
Questions
Can it be used for custom voice cloning?
From 2 comments
What I am still looking for is a way to clone voice locally. I have OK hardware. For example I can use Mistral Small 3.1 or what it is called locally. Premade voices can be…
zelphirkalt on HNI wonder what would it take to extend it with a custom voice?
BenGosub on HN
How does it compare to Piper?
From 1 comment
How does this compare to say piper-tts? I ask because their models are pretty small. Some sound awesome and there is no depdendency hell like I'm seeing here. Example:…
77pt77 on HN
Alternatives named
Piper TTS
From 2 comments
How does this compare to say piper-tts? I ask because their models are pretty small. Some sound awesome and there is no depdendency hell like I'm seeing here. Example:…
77pt77 on HNAwesome work! Often times in the TTS space, human-similarity is given way too much emphasis at the expense of hurting user access. Frankly as long as a voice is clear and you…
C-Loftus on HN
Kokoro TTS
From 2 comments
Hmm the quality is not so impressive. I'm looking for a really naturally sounding model. Not very happy with piper/kokoro, XTTS was a bit complex to set up. For STT whisper is…
wkat4242 on HNamazing! can't wait to integrate it into https://desktop.with.audio I'm already using KokorosTTS without a GPU. It works fairly well on Apple Silicon. Foundational tools like…
OfflineSergio on HN
F5-TTS and Fish-Speech
From 1 comment
Cool. While I think this is indeed impressive and has a specific use case (e.g. in the embedded sector), I'm not totally convinced that the quality is good enough to replace…
sandreas on HN