Gencho: Room Impulse Response Generation from Reverberant Speech and Text via Diffusion Transformers

2026-02-09

AI Summary

This work proposes Gencho, a diffusion Transformer-based model for blind deconvolution in unknown acoustic environments, addressing the limited modeling capacity of existing methods and their incompatibility with generative audio applications. Gencho employs a structure-aware encoder to disentangle early and late reflection components from reverberant speech and predicts room impulse responses (RIRs) in complex spectrogram form. By incorporating text-conditioned control, the model enables diverse and perceptually realistic acoustic environment generation. Gencho supports modular integration and text-guided synthesis, outperforming non-generative baselines in both subjective listening quality and objective metrics, and demonstrates successful application in text-driven acoustic simulation tasks.

Technology Category

• Natural Language Processing: Generation
• Machine Learning: Deep Generative Models & Autoencoders
• Computer Vision: Diffusion Models for Vision

Application Category

• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Search and Retrieval-Augmented AI: Large language models for search
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

Blind room impulse response (RIR) estimation is a core task for capturing and transferring acoustic properties; yet existing methods often suffer from limited modeling capability and degraded performance under unseen conditions. Moreover, emerging generative audio applications call for more flexible impulse response generation methods. We propose Gencho, a diffusion-transformer-based model that predicts complex spectrogram RIRs from reverberant speech. A structure-aware encoder leverages isolation between early and late reflections to encode the input audio into a robust representation for conditioning, while the diffusion decoder generates diverse and perceptually realistic impulse responses from it. Gencho integrates modularly with standard speech processing pipelines for acoustic matching. Results show richer generated RIRs than non-generative baselines while maintaining strong performance in standard RIR metrics. We further demonstrate its application to text-conditioned RIR generation, highlighting Gencho's versatility for controllable acoustic simulation and generative audio tasks.

Problem

• Room Impulse Response
• Blind Estimation
• Generative Audio
• Acoustic Simulation
• Reverberant Speech

Innovation

• Diffusion Transformer
• Room Impulse Response (RIR) generation
• Reverberant speech
• Text-conditioned audio synthesis
• Structure-aware encoding