Open Models Almanac
Methodsmodel architecturemultimodal architecture

single mention · model architecture

Audio Transformer (AuT)

AuT is a transformer-based audio encoder trained from scratch within an attention-encoder-decoder model.1 It downsamples audio filter bank features with Conv2D blocks and feeds them into self-attention layers, producing audio tokens at a fixed token rate.2

How this definition was written

Summarised by glm-5.3-flash from 1 source; every sentence is checked against the quote it cites.

  1. We use transformer based audio encoder trained from scratch in attention-encoder-decoder model AuT — Qwen3.5-Omni Technical Report
  2. The filter bank features of the audio are downsampled 16 times using 4 Conv2D blocks and then fed into self-attention layers to obtain audio tokens — Qwen3.5-Omni Technical Report

Also called attention-encoder-decoder model AuT.

Filed under model architecture › multimodal architecture