Support streaming with timestamps

[gad2103]

Hello and thank you for this nice rust wrapper for parakeet.

I'm wondering if it's possible to add support for streaming with timestamps. As far as I understand this requires using the TDT model family.

If it's pretty straightforward I'm happy to take on the PR work as I really need this for a project I'm working on and don't want to have to resort to using python over the network or even a local socket. Let me know!

[altunenes]

Hello, current exposed output is strictly the transcribed text and the signal that the speaker has finished speaking (< EOU > ).
could you please try this in offical version in python first to verify if the results are actually usable? because I remember that there was no such output in the original version either. note that this model is strictly optimized for latency and EOU detection rather than alignment, but we can technically derive word-level timestamps from frame indices, but without punctuation support, I am pessimistic about how clean/good the output will actually look :/

[gad2103]

I'll take a look at the python version to see what happens