93 comments

[ 4.3 ms ] story [ 77.1 ms ] thread
> Apple M2 Pro: 4.5 seconds

> AMD Ryzen 7 8745HS: 1.5 seconds

These two can probably do it much faster on their iGPUs.

Yep, and Kokoro-FastAPI (which he already uses) makes it super easy with start-gpu_mac.sh
kokoro is decent but pocket-tts is much better especially when you rip a good voice. https://github.com/kyutai-labs/pocket-tts

the onnx version of pocket-tts does perform better. https://huggingface.co/KevinAHM/pocket-tts-onnx

I've found that for CPU inference the PyTorch-based (non-quantized) version of Pocket TTS actually performs (both speed and quality-wise) better than the ONNX version, even after fiddling with all of the knobs that ONNX provides.
I agree, pocket-tts is quite good. It is also very easy to make new voices from small sample files with pocket-tts.
I'm using exactly this TTS engine for my intercom door system I built. The quality of the TTS is very good.
kokoro is surprisingly great at nuance but it's tough to improve that last ~2% or so. kokoro + rvc is really great too; i use that for ELEMENT47, the LLM-centric comedy podcast i do that i wish more people would listen to. (e47.net , feel free to subscribe!)
Any good debian-ish distros that integrate TTS and STT in a usable shell?
kokoro is very nice, but I am disappointed that this wasn't an announcement of a new kokoro version.
I'm using Kokoro for a fun little side-project browser-based game I'm working on. It's legitimately super good for being only 85mb (for the wasm version) or 300mb (for the webgpu version).
Fun... This is something I actually care about...

I used to keep a version of whisperx around, because I think it's important to have not just transcription, but also timing and speaker identification (e.g. for subtitles)... It depends on pyannote, though, which has some wierd licensing (and is tougher to script the installs because of it), so I wanted to look at something that both had better transcription, and supported diarization (the speaker and timing). I decided on parakeet for the transcription with softformer (the diarization), but most of the available engines for it don't include softformer.

I coded up an OpenAI compatible server for parakeet-rs ( https://github.com/altunenes/parakeet-rs ) (which does support softformer) and I've been using it with OpenWhispr (a desktop app for transcription that handles all sorts of neat thing).

I'm doing CPU-only transcription (because I use my GPUs for other stuff and haven't gotten around to adding in the GPU-path), but it's incredibly empowering to be able to have local transcriptions at will.

This is TTS. Not STT.

For what you are doing, Senko works really well for diarization along with parakeet.

Faster and more accurate than Pyannote and whisper on my MacBook anyway.

You're right... I read the title too quickly... I'll have to look at Senko vs Softformer later...
Still I for one am grateful for any STT transcription models.
Yeah, we need to keep up with how quickly AI types back to us, typing on the keyboards is no longer quick enough, gotta dictate everything now.
Great way to enter your passwords
Cool I actually got it ported to iPhone’s ANE finally yesterday! So we can get both rt natural local TTS and 4x less battery drainage and thermals
lol, I just finished a port to the iPhone ANE as well. I would love to compare notes
The biggest part was dynamic shapes that weren’t allowed for newer iOS. I’d do it with masking similar to other transformers being ported to ANE. Another thing was people would cut the graph cause some operations are not allowable on this chip - tho if you look at the op that is doing segfault, i.e. tile - you can rewrite the same functionality with other supported ops and port the weights to the new graph. Works the same way compared to PyTorch weights comparing spectrograms, around 10x realtime on 14pro :)

Anything you found interesting there? Every proposal on GitHub I found was kinda lucking in both documentation and completeness

I have used Kokoro fairly extensively for an accessibility product. I have loved working with it (especially because I don't have an NVidia GPU like many TTS of similar quality require).

I particularly appreciate the fact that it lets you manually add IPA pronunciation guides. There have been some cases where an important word is a homograph and Kokoro assumed the wrong pronunciation.

The place where it falls a little short is in saying just a single word or two. Try having it say simply "six" and it almost always says something like "ah-six-ah". I found a way around that though. If you give it a longer sentence to say (eg "The word is: six") it will say it fine. The trick is that the Kokoro API gives you the timestamp of each word in the sentence. So you can have a Python script crop out just the word you care about. The intonation is a little flat this way, but is very reliable.

I asked about this on the discord, and was told that it is a limitation of the small parameter size. But in fairness to Kokoro, even eleven-labs' voices suffer from this occasionally.

Great technique, thanks for sharing
Same trick with many others. I say "Knight" to Wispr or Google and see "night". I say "Knight to f3", and just backspace out "to f3".
Of course, over time, I use custom entries.

My snippet expansion entry in Wispr is "Chess Knight" = "Knight" ("Knight to f3" without customization was more reliable than "Chess Knight")

I also use "dot bullet" = "•", as I like to separate thoughts with • more than ;

It is indeed a common weakness of TTS models.

Unfortunately it makes it unsuited for my use case, which is almost entirely single words, as I don't particularly want to deal with stitching/segmenting input/output.

Love Kokoro tts. I wrote https://github.com/Jud/kokoro-coreml to try pushing the limits a bit on speed & size. Such great quality at a given size. As others have mentioned short utterances are problematic, but solvable.
Anyone know which local TTS is best, close to Eleven Labs quality?
F5-TTS
(comment deleted)
curious to know if it comes with audio tags?
Both Text-to-Speech and Speech-to-Text now have local models that are good enough to get the job done. Kokoro for TTS, Parakeet for STT and Fluid-1 for text formatting (I use it with FluidVoice). I hope this is a trend that continues for other applications.
I use kokoro with home assistant and its great. I find its the most natural sounding and small too.

I speak over sonos speakers when certain events happen. And use it as my voice assistant.

It's interesting that the male voices are all so much worse than the female voices (several are quite good). There is bias in machine learning, but I wonder whether there is also systematically more training data of female speech?
Love this model. I’m GPU poor and have had FOMO that I haven’t played with local models at all. About a month ago I setup Kokoro on my GTX1650 to do TTS for an article reader. A simple WebUI lets me paste a URL or a chunk of copy pasted text. Python cleans it up and sends to Kokoro for TTS and it’s then served via RSS for Apple Podcasts. Then for my morning drive I’ll catch up on articles or blog posts I’ve gathered.

At some point I’d like to play with separate voices and see if I could build something like NotebookLM for kind of like a radio morning show of news items I’ve gathered.

When I hear the male voice I think: "Ok, it's the Youtube guy".
I just hooked it up to my personal AI Japanese Teacher app, pretty good quality / natural sounding speech in mixed English / Japanese while running fast on CPU so I don't waste VRAM.
Was just thinking of making exactly that. I have made Kokoro the voice of my local LLM (qwen 3.6). What do you use for a Japanese tutor app?