What does `non-continuous` recognition means for an implementationt that is only local

[padenot]

It seem that it was intended for high-latency systems, I'm not sure how it has to work when that isn't the case.

[evanbliu]

The spec says this about the continuous attribute:

When the continuous attribute is set to false, the user agent must return no more than one final result in response to starting recognition, for example a single turn pattern of interaction. When the continuous attribute is set to true, the user agent must return zero or more final results representing multiple consecutive recognitions in response to starting recognition, for example a dictation. The default value must be false. Note, this attribute setting does not affect interim results.

While this may have been intended for high-latency systems, I guess a low-latency system could still technically follow this behavior. We can propose a change to the spec if you'd like, but it might be simpler to just keep it as is.

[padenot]

Yeah I was simply wondering what is supposed to happen if I set this to non-continuous, then proceed to feed 30 minutes of audio. Per spec we want one callback with 30min worth of transcription, but maybe interim results are allowed.

[evanbliu]

Oh I see. Yeah, it looks like the spec doesn't mention any restrictions on the number of interim results when continuous is set to false.