喫煙する人が一人もいません - it reads really fast so I had to listen a few times, but thanks to it I didn’t have to guess which reading it was.
Fun.
3 Likes
Have you tried Azure neural text-to-speech?
3 Likes
Deepl just says “Chinese Letter” over and over in an incredibly robotic voice for me… is there a secret?
The Azure one is really good though
2 Likes
I chose translate from Japanese
My Deepl doesn’t seem to have a voice function at all for Japanese. Is that a paid feature?
At any rate, I just use Google Translate or Ichi.moe if I’m unsure of the reading.
I missed the n at the end. It was fun. I was taking away letters from the end, one by one, and then I took away some middle letters - the meaning changed but the translator always gave a result
1 Like
Huh. I’ve never noticed that “read it” button before, and I’ve definitely looked for it…
2 Likes
DeepL’s strength and weakness is that it always tries to output something sensible in the target language. This makes it useful for translating content produced by a native speaker where you can assume that the grammar is reasonably correct. It will produce idiomatically acceptable English rather than a stilted, literal translation.
It also makes it a very bad tool for checking whether or not you’ve gotten your own grammar right because you can write some seriously janky, broken Japanese and it will still output a coherent English sentence rather than an English sentence that is equivalently janky to your Japanese input.
2 Likes
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.