termux-speech-to-text buffers progressive output

[yesco]

Problem description
termux-speech-to-text -p

doesn't print progressive output as it comes, but seems to buffer it. It comes out in a larger chunk.

Steps to reproduce
Run the command

termux-speech-to-text -p

speak a lot, watch as no output comes until either the end, or when enough has buffered up.

Expected behavior
I'd expect it to output each line as each new word is recognized. That's what progressive means.

Additional information
I've tried to see if it's the terminal or script that buffers, so I wrapped the execution in:

stdbuf -i0 -o0 /data/data/com.termux/files/usr/libexec/termux-api SpeechToText

But there is no difference.