๐ฎ
๐ฎ
The Ethereal
Parallel Time-Band Mixing with Learned Observation-Adding for Robust ASR Front-Ends
August 31, 2026 ยท Grace Period ยท ๐ Interspeech 2026
Authors
Xingyu Shen, Runze Wang, Wei-Ping Zhu, Benoit Champagne
arXiv ID
2608.30326
Category
cs.SD: Sound
Cross-listed
cs.AI
Citations
0
Venue
Interspeech 2026
Abstract
Speech enhancement is often used as a front-end for robust ASR, yet recurrent temporal and cross-band modules introduce sequential dependencies that reduce parallel efficiency. In this paper, we present a sequence-parallel band-split enhancement front-end built on a Parallel Time-Band Mixer (PTBM) block that eliminates within-block recurrent unrolling. PTBM integrates intra-band temporal mixing and per-frame cross-band attention within a unified parallel architecture, enabling efficient contextual modeling across both time and frequency dimensions. The system retains the mask-plus-residual reconstruction interface and introduces learned Observation-Adding (LOA) to suppress ASR-sensitive artifacts without development-set tuning. Experiments on DNS Challenge and CHiME-4 with frozen Whisper back-ends show that the proposed front-end consistently reduces word error rate relative to recurrent band-split baselines while requiring only 0.96 M parameters and 0.58 GMAC/s for the front-end network.
Community Contributions
Found the code? Know the venue? Think something is wrong? Let us know!
๐ Similar Papers
In the same crypt โ Sound
R.I.P.
๐ป
Ghosted
Multi-talker Speech Separation with Utterance-level Permutation Invariant Training of Deep Recurrent Neural Networks
R.I.P.
๐ป
Ghosted
The fifth 'CHiME' Speech Separation and Recognition Challenge: Dataset, task and baselines
R.I.P.
๐ป
Ghosted
TasNet: time-domain audio separation network for real-time, single-channel speech separation
R.I.P.
๐ป
Ghosted
SampleRNN: An Unconditional End-to-End Neural Audio Generation Model
R.I.P.
๐ป
Ghosted