AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

MFlowAudio: Efficient Text-to-Audio Synthesis via Mamba-based Stateful Flow Matching

Hao Dai, Panyu Chen, Jagmohan Chauhan

NeurIPS 2026 · Image generation › Autoregressive and latent generation · approach: Diffusion audio synthesis · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes MFlowAudio, a latent text-to-audio generator that combines flow matching with a custom Mamba-based (TFMamba) backbone to avoid the quadratic cost of self-attention in long-form audio synthesis.

Problem

Transformer-based diffusion models for audio generation suffer from quadratic self-attention complexity, which bottlenecks long-form text-to-audio synthesis.

Approach

Pairs continuous-time flow matching in latent space with a custom TFMamba block (state-space model) and a stateful design, replacing self-attention to achieve efficient long-form audio generation.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace Transformer backbone with TCMamba (Mamba-based) under similar sizeMFlowAudio-base vs TangoFlux-base · AudioCaps 10sFD: 16.9351 -> 10.9178Improved
use TCMamba instead of TransformerMFlowAudio vs TangoFlux · AudioCaps 10sCLAP and IS: IS 12.5489 -> 11.6865; CLAP 0.4990 -> 0.4957Hurt
use Mamba backbone with adaptive-length computeMFlowAudio · generation latency 10slatency (s): 3.51 -> 1.93Improved
add state propagation via SFM chunked generationMFlowAudio · AudioCapsFD: 11.8569 -> 11.5988 (chunk 64)No clear effect
remove state propagation (overlap-add) in prompt switching (ablation)MFlowAudio · dynamic prompt switchingBBR: 4.3287 -> 7.4001Hurt
add SDCG vs vanilla conditional field (ablation)MFlowAudio · dynamic prompt switchingFD: 16.0940 -> 15.8575Improved
add SDCG vs vanilla conditional field (ablation)MFlowAudio · dynamic prompt switchingBBR: 4.2633 -> 4.3287Hurt
add SDCG vs naive subtraction (ablation)MFlowAudio · dynamic prompt switchingFD: 16.0056 -> 15.8575Improved
add stateful dynamic prompting vs TangoFlux stateless prompt switchingMFlowAudio · dynamic prompt switchingBBR: 11.6443 -> 4.3287Improved
use 8-frame overlap in VAE chunk decoding (ablation)Oobleck VAE · chunk decodingrel-L1: 4.54% -> 0.049%Improved
increase overlap from 8 to 16 frames (ablation)Oobleck VAE · chunk decodingrel-L1: 0.049% -> 0.047%No clear effect
stream with chunk size 64 vs one-time generation, 30s batch 2MFlowAudio · latencytime (s): 5.1 -> 1.85Improved

Experimental setup

Builds on: flow matching Mamba

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy