single mention · not yet filed
Adaptive Rate Interleave Alignment (ARIA)
ARIA unifies the conventional dual-channel generation paradigm into a single-channel formulation for text and speech tokens.1 It enforces an adaptive rate constraint where the cumulative speech-to-text token ratio must not exceed the corresponding item-level global ratio.2
Trade-off ARIA reduces the synchronization overhead between separate generation tracks and enables more efficient token scheduling during decoding.3
How this definition was written
Summarised by glm-5.3-flash from 1 source; every sentence is checked against the quote it cites.
we propose ARIA (Adaptive Rate Interleave Alignment), which unifies the conventional dual-channel generation paradigm into a single-channel formulation
— Qwen3.5-Omni Technical ReportARIA enforces an adaptive rate constraint: for any prefix of the generated sequence, the cumulative speech-to-text token ratio must not exceed the corresponding item-level global ratio
— Qwen3.5-Omni Technical ReportARIA reduces the synchronization overhead between separate generation tracks, enables more efficient token scheduling during decoding, and better matches the naturally incremental regime of streaming interaction
— Qwen3.5-Omni Technical Report
Also called ARIA (Adaptive Rate Interleave Alignment).