Was playing around a bit and for its size it's very impressive. Just has issues pronounciating numbers. I tried to let it generate "Startup finished in 135 ms."
I didn't expect it to pronounciate 'ms' correctly, but the number sounded just like noise. Eventually I got an acceptable result for the string "Startup finished in one hundred and thirty five seconds.
yeah we're fixing this at the model level too. but in the meantime, there is a way to add text preprocessing for you, and if you have a special use-cased, claude code should be able to one-shot custom preprocessing. its the way that most existing tts models (including sota cloud ones) deal w numbers and units, they just convert it into string.
thanks a lot for trying it and giving feedback. custom preprocessing will fix this for 95% of use-cases. and as i mentioned, this will be fixed at the model level in the next release.
The above SECDED check-bit encoding can be implemented in a similar way, but since it uses only three-bit patterns, mapping syndromes to correction masks can be done with three-input AND gates.
It sounded quite good indeed for the normal English stuff, but I guess predictably was quite bad at the domain-specific words. It misspoke "SECDED", had wrong emphasis on "syndromes", and pronounced "AND gates" like "and gates".
Could you give some example of what kind of preprocessing would help in this case? I tried some local LLMs, but they didn't do a good job (maybe my prompts sucked).
I'm not sure if you're misspelling it deliberately or not, but the word you're looking for is "pronounce" and it's verb form "pronouncing", as in "It just has issues pronouncing numbers" and "I didn't expect it to pronounce 'ms' correctly."
- d2 is a standalone executable compiler, I once tried mermaid-cli (mmdc) but couldn't get it to work properly plus anything I need to install with npm scares the hell out of me
- ASCII rendering: I love rendering to ASCII which I can copy-paste around.
But I do use mermaid a lot embedded in other programs (e.g Obisidian). The selection of different diagram types is amazing.
I've been fully on DDG for years but becoming slowly skeptical & looking for alternatives.
1. They're leaning heavily into "responsible AI", much like Mozilla
2. Might be just me but I feel like their algorithm became significantly worse recently. Over the years they've gone from being worse than Google in the early days to steadily improving & overtaking Google on quality (I made heavy use of !g until I started slowly realising it was no longer giving me better results). But now I feel like they've reversed & regressed again.
Brilliant. Really nice looking TUI.
One thing I noticed is that I still find myself using the mouse to click the form fields. The keyboard navigation seems to sometimes get stuck on fields and I then can't move around anymore. Is there an easy trick for jumping between the fields?
It's interesting how the level of public computer/computing knowledge changed.
The Byte magazine goes into deep details of hardware, software and programming.
I feel that nowadays a lot of it is taking for granted or very few people care how things work under the hood. But probably at the time of the Byte magazine only very few people cared too :-).
I also noticed that they struggle reversing strings.
Ask it to "generate a list of the 30 biggest countries together with their name in reverse". Most of the results will be correct but you'll likely find some weird spelling mistakes.
It's not something they can regurgitate from previously seen text. Models like Claude with background code execution might get around that.
reply