Deferred NAM: Low-latency Top-K Context Injection via Deferred Context Encoding for Non-Streaming ASR

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wu, Zelin, Song, Gan, Li, Christopher, Rondon, Pat, Meng, Zhong, Velez, Xavier, Wang, Weiran, Caseiro, Diamantino, Pundak, Golan, Munkhdalai, Tsendsuren, Chandorkar, Angad, Prabhavalkar, Rohit
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913325797867520
author Wu, Zelin
Song, Gan
Li, Christopher
Rondon, Pat
Meng, Zhong
Velez, Xavier
Wang, Weiran
Caseiro, Diamantino
Pundak, Golan
Munkhdalai, Tsendsuren
Chandorkar, Angad
Prabhavalkar, Rohit
author_facet Wu, Zelin
Song, Gan
Li, Christopher
Rondon, Pat
Meng, Zhong
Velez, Xavier
Wang, Weiran
Caseiro, Diamantino
Pundak, Golan
Munkhdalai, Tsendsuren
Chandorkar, Angad
Prabhavalkar, Rohit
contents Contextual biasing enables speech recognizers to transcribe important phrases in the speaker's context, such as contact names, even if they are rare in, or absent from, the training data. Attention-based biasing is a leading approach which allows for full end-to-end cotraining of the recognizer and biasing system and requires no separate inference-time components. Such biasers typically consist of a context encoder; followed by a context filter which narrows down the context to apply, improving per-step inference time; and, finally, context application via cross attention. Though much work has gone into optimizing per-frame performance, the context encoder is at least as important: recognition cannot begin before context encoding ends. Here, we show the lightweight phrase selection pass can be moved before context encoding, resulting in a speedup of up to 16.1 times and enabling biasing to scale to 20K phrases with a maximum pre-decoding delay under 33ms. With the addition of phrase- and wordpiece-level cross-entropy losses, our technique also achieves up to a 37.5% relative WER reduction over the baseline without the losses and lightweight phrase selection pass.
format Preprint
id arxiv_https___arxiv_org_abs_2404_10180
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Deferred NAM: Low-latency Top-K Context Injection via Deferred Context Encoding for Non-Streaming ASR
Wu, Zelin
Song, Gan
Li, Christopher
Rondon, Pat
Meng, Zhong
Velez, Xavier
Wang, Weiran
Caseiro, Diamantino
Pundak, Golan
Munkhdalai, Tsendsuren
Chandorkar, Angad
Prabhavalkar, Rohit
Computation and Language
Artificial Intelligence
Machine Learning
Neural and Evolutionary Computing
Audio and Speech Processing
Contextual biasing enables speech recognizers to transcribe important phrases in the speaker's context, such as contact names, even if they are rare in, or absent from, the training data. Attention-based biasing is a leading approach which allows for full end-to-end cotraining of the recognizer and biasing system and requires no separate inference-time components. Such biasers typically consist of a context encoder; followed by a context filter which narrows down the context to apply, improving per-step inference time; and, finally, context application via cross attention. Though much work has gone into optimizing per-frame performance, the context encoder is at least as important: recognition cannot begin before context encoding ends. Here, we show the lightweight phrase selection pass can be moved before context encoding, resulting in a speedup of up to 16.1 times and enabling biasing to scale to 20K phrases with a maximum pre-decoding delay under 33ms. With the addition of phrase- and wordpiece-level cross-entropy losses, our technique also achieves up to a 37.5% relative WER reduction over the baseline without the losses and lightweight phrase selection pass.
title Deferred NAM: Low-latency Top-K Context Injection via Deferred Context Encoding for Non-Streaming ASR
topic Computation and Language
Artificial Intelligence
Machine Learning
Neural and Evolutionary Computing
Audio and Speech Processing
url https://arxiv.org/abs/2404.10180