[dismantlethesun]

> It's because each syllable has to have its own ideograph (usually there are multiple ideographs) and you can't easily learn 10000 of them

As an aside, one side-benefit of a language that doesn't have so many phenomes is that automatic speech creation is much easier.

That's speech creation, not speech recognition.

Using software to speak Japanese fluently, at a given pitch, with a given emotional context is a solved problem. There's even software that can be used to create singing voices.

It only takes about 1000 recorded sounds to do so. In english, it'd take 10,000 multiplied by the number of emotional contexts and speech patterns you'd like to mimic.