Alert button
Picture for Jay Mahadeokar

Jay Mahadeokar

Alert button

TODM: Train Once Deploy Many Efficient Supernet-Based RNN-T Compression For On-device ASR Models

Sep 05, 2023
Yuan Shangguan, Haichuan Yang, Danni Li, Chunyang Wu, Yassir Fathullah, Dilin Wang, Ayushi Dalmia, Raghuraman Krishnamoorthi, Ozlem Kalinli, Junteng Jia, Jay Mahadeokar, Xin Lei, Mike Seltzer, Vikas Chandra

Figure 1 for TODM: Train Once Deploy Many Efficient Supernet-Based RNN-T Compression For On-device ASR Models
Figure 2 for TODM: Train Once Deploy Many Efficient Supernet-Based RNN-T Compression For On-device ASR Models
Figure 3 for TODM: Train Once Deploy Many Efficient Supernet-Based RNN-T Compression For On-device ASR Models
Figure 4 for TODM: Train Once Deploy Many Efficient Supernet-Based RNN-T Compression For On-device ASR Models

Automatic Speech Recognition (ASR) models need to be optimized for specific hardware before they can be deployed on devices. This can be done by tuning the model's hyperparameters or exploring variations in its architecture. Re-training and re-validating models after making these changes can be a resource-intensive task. This paper presents TODM (Train Once Deploy Many), a new approach to efficiently train many sizes of hardware-friendly on-device ASR models with comparable GPU-hours to that of a single training job. TODM leverages insights from prior work on Supernet, where Recurrent Neural Network Transducer (RNN-T) models share weights within a Supernet. It reduces layer sizes and widths of the Supernet to obtain subnetworks, making them smaller models suitable for all hardware types. We introduce a novel combination of three techniques to improve the outcomes of the TODM Supernet: adaptive dropouts, an in-place Alpha-divergence knowledge distillation, and the use of ScaledAdam optimizer. We validate our approach by comparing Supernet-trained versus individually tuned Multi-Head State Space Model (MH-SSM) RNN-T using LibriSpeech. Results demonstrate that our TODM Supernet either matches or surpasses the performance of manually tuned models by up to a relative of 3% better in word error rate (WER), while efficiently keeping the cost of training many models at a small constant.

* Meta AI; Submitted to ICASSP 2024 
Viaarxiv icon

Prompting Large Language Models with Speech Recognition Abilities

Jul 21, 2023
Yassir Fathullah, Chunyang Wu, Egor Lakomkin, Junteng Jia, Yuan Shangguan, Ke Li, Jinxi Guo, Wenhan Xiong, Jay Mahadeokar, Ozlem Kalinli, Christian Fuegen, Mike Seltzer

Figure 1 for Prompting Large Language Models with Speech Recognition Abilities
Figure 2 for Prompting Large Language Models with Speech Recognition Abilities
Figure 3 for Prompting Large Language Models with Speech Recognition Abilities
Figure 4 for Prompting Large Language Models with Speech Recognition Abilities

Large language models have proven themselves highly flexible, able to solve a wide range of generative tasks, such as abstractive summarization and open-ended question answering. In this paper we extend the capabilities of LLMs by directly attaching a small audio encoder allowing it to perform speech recognition. By directly prepending a sequence of audial embeddings to the text token embeddings, the LLM can be converted to an automatic speech recognition (ASR) system, and be used in the exact same manner as its textual counterpart. Experiments on Multilingual LibriSpeech (MLS) show that incorporating a conformer encoder into the open sourced LLaMA-7B allows it to outperform monolingual baselines by 18% and perform multilingual speech recognition despite LLaMA being trained overwhelmingly on English text. Furthermore, we perform ablation studies to investigate whether the LLM can be completely frozen during training to maintain its original capabilities, scaling up the audio encoder, and increasing the audio encoder striding to generate fewer embeddings. The results from these studies show that multilingual ASR is possible even when the LLM is frozen or when strides of almost 1 second are used in the audio encoder opening up the possibility for LLMs to operate on long-form audio.

Viaarxiv icon

Voicebox: Text-Guided Multilingual Universal Speech Generation at Scale

Jun 23, 2023
Matthew Le, Apoorv Vyas, Bowen Shi, Brian Karrer, Leda Sari, Rashel Moritz, Mary Williamson, Vimal Manohar, Yossi Adi, Jay Mahadeokar, Wei-Ning Hsu

Figure 1 for Voicebox: Text-Guided Multilingual Universal Speech Generation at Scale
Figure 2 for Voicebox: Text-Guided Multilingual Universal Speech Generation at Scale
Figure 3 for Voicebox: Text-Guided Multilingual Universal Speech Generation at Scale
Figure 4 for Voicebox: Text-Guided Multilingual Universal Speech Generation at Scale

Large-scale generative models such as GPT and DALL-E have revolutionized natural language processing and computer vision research. These models not only generate high fidelity text or image outputs, but are also generalists which can solve tasks not explicitly taught. In contrast, speech generative models are still primitive in terms of scale and task generalization. In this paper, we present Voicebox, the most versatile text-guided generative model for speech at scale. Voicebox is a non-autoregressive flow-matching model trained to infill speech, given audio context and text, trained on over 50K hours of speech that are neither filtered nor enhanced. Similar to GPT, Voicebox can perform many different tasks through in-context learning, but is more flexible as it can also condition on future context. Voicebox can be used for mono or cross-lingual zero-shot text-to-speech synthesis, noise removal, content editing, style conversion, and diverse sample generation. In particular, Voicebox outperforms the state-of-the-art zero-shot TTS model VALL-E on both intelligibility (5.9% vs 1.9% word error rates) and audio similarity (0.580 vs 0.681) while being up to 20 times faster. See voicebox.metademolab.com for a demo of the model.

Viaarxiv icon

Towards Selection of Text-to-speech Data to Augment ASR Training

May 30, 2023
Shuo Liu, Leda Sarı, Chunyang Wu, Gil Keren, Yuan Shangguan, Jay Mahadeokar, Ozlem Kalinli

Figure 1 for Towards Selection of Text-to-speech Data to Augment ASR Training
Figure 2 for Towards Selection of Text-to-speech Data to Augment ASR Training
Figure 3 for Towards Selection of Text-to-speech Data to Augment ASR Training
Figure 4 for Towards Selection of Text-to-speech Data to Augment ASR Training

This paper presents a method for selecting appropriate synthetic speech samples from a given large text-to-speech (TTS) dataset as supplementary training data for an automatic speech recognition (ASR) model. We trained a neural network, which can be optimised using cross-entropy loss or Arcface loss, to measure the similarity of a synthetic data to real speech. We found that incorporating synthetic samples with considerable dissimilarity to real speech, owing in part to lexical differences, into ASR training is crucial for boosting recognition performance. Experimental results on Librispeech test sets indicate that, in order to maintain the same speech recognition accuracy as when using all TTS data, our proposed solution can reduce the size of the TTS data down below its $30\,\%$, which is superior to several baseline methods.

Viaarxiv icon

Multi-Head State Space Model for Speech Recognition

May 25, 2023
Yassir Fathullah, Chunyang Wu, Yuan Shangguan, Junteng Jia, Wenhan Xiong, Jay Mahadeokar, Chunxi Liu, Yangyang Shi, Ozlem Kalinli, Mike Seltzer, Mark J. F. Gales

Figure 1 for Multi-Head State Space Model for Speech Recognition
Figure 2 for Multi-Head State Space Model for Speech Recognition
Figure 3 for Multi-Head State Space Model for Speech Recognition
Figure 4 for Multi-Head State Space Model for Speech Recognition

State space models (SSMs) have recently shown promising results on small-scale sequence and language modelling tasks, rivalling and outperforming many attention-based approaches. In this paper, we propose a multi-head state space (MH-SSM) architecture equipped with special gating mechanisms, where parallel heads are taught to learn local and global temporal dynamics on sequence data. As a drop-in replacement for multi-head attention in transformer encoders, this new model significantly outperforms the transformer transducer on the LibriSpeech speech recognition corpus. Furthermore, we augment the transformer block with MH-SSMs layers, referred to as the Stateformer, achieving state-of-the-art performance on the LibriSpeech task, with word error rates of 1.76\%/4.37\% on the development and 1.91\%/4.36\% on the test sets without using an external language model.

* Interspeech 2023 
Viaarxiv icon

Improving Fast-slow Encoder based Transducer with Streaming Deliberation

Dec 15, 2022
Ke Li, Jay Mahadeokar, Jinxi Guo, Yangyang Shi, Gil Keren, Ozlem Kalinli, Michael L. Seltzer, Duc Le

Figure 1 for Improving Fast-slow Encoder based Transducer with Streaming Deliberation
Figure 2 for Improving Fast-slow Encoder based Transducer with Streaming Deliberation
Figure 3 for Improving Fast-slow Encoder based Transducer with Streaming Deliberation
Figure 4 for Improving Fast-slow Encoder based Transducer with Streaming Deliberation

This paper introduces a fast-slow encoder based transducer with streaming deliberation for end-to-end automatic speech recognition. We aim to improve the recognition accuracy of the fast-slow encoder based transducer while keeping its latency low by integrating a streaming deliberation model. Specifically, the deliberation model leverages partial hypotheses from the streaming fast encoder and implicitly learns to correct recognition errors. We modify the parallel beam search algorithm for fast-slow encoder based transducer to be efficient and compatible with the deliberation model. In addition, the deliberation model is designed to process streaming data. To further improve the deliberation performance, a simple text augmentation approach is explored. We also compare LSTM and Conformer models for encoding partial hypotheses. Experiments on Librispeech and in-house data show relative WER reductions (WERRs) from 3% to 5% with a slight increase in model size and negligible extra token emission latency compared with fast-slow encoder based transducer. Compared with vanilla neural transducers, the proposed deliberation model together with fast-slow encoder based transducer obtains relative 10-11% WERRs on Librispeech and around relative 6% WERR on in-house data with smaller emission delays.

* Submitted to ICASSP 2023 
Viaarxiv icon

Dynamic Speech Endpoint Detection with Regression Targets

Oct 25, 2022
Dawei Liang, Hang Su, Tarun Singh, Jay Mahadeokar, Shanil Puri, Jiedan Zhu, Edison Thomaz, Mike Seltzer

Figure 1 for Dynamic Speech Endpoint Detection with Regression Targets
Figure 2 for Dynamic Speech Endpoint Detection with Regression Targets
Figure 3 for Dynamic Speech Endpoint Detection with Regression Targets
Figure 4 for Dynamic Speech Endpoint Detection with Regression Targets

Interactive voice assistants have been widely used as input interfaces in various scenarios, e.g. on smart homes devices, wearables and on AR devices. Detecting the end of a speech query, i.e. speech end-pointing, is an important task for voice assistants to interact with users. Traditionally, speech end-pointing is based on pure classification methods along with arbitrary binary targets. In this paper, we propose a novel regression-based speech end-pointing model, which enables an end-pointer to adjust its detection behavior based on context of user queries. Specifically, we present a pause modeling method and show its effectiveness for dynamic end-pointing. Based on our experiments with vendor-collected smartphone and wearables speech queries, our strategy shows a better trade-off between endpointing latency and accuracy, compared to the traditional classification-based method. We further discuss the benefits of this model and generalization of the framework in the paper.

* Manuscript submitted to ICASSP 2023 
Viaarxiv icon

An Investigation of Monotonic Transducers for Large-Scale Automatic Speech Recognition

Apr 19, 2022
Niko Moritz, Frank Seide, Duc Le, Jay Mahadeokar, Christian Fuegen

Figure 1 for An Investigation of Monotonic Transducers for Large-Scale Automatic Speech Recognition
Figure 2 for An Investigation of Monotonic Transducers for Large-Scale Automatic Speech Recognition
Figure 3 for An Investigation of Monotonic Transducers for Large-Scale Automatic Speech Recognition
Figure 4 for An Investigation of Monotonic Transducers for Large-Scale Automatic Speech Recognition

The two most popular loss functions for streaming end-to-end automatic speech recognition (ASR) are the RNN-Transducer (RNN-T) and the connectionist temporal classification (CTC) objectives. Both perform an alignment-free training by marginalizing over all possible alignments, but use different transition rules. Between these two loss types we can classify the monotonic RNN-T (MonoRNN-T) and the recently proposed CTC-like Transducer (CTC-T), which both can be realized using the graph temporal classification-transducer (GTC-T) loss function. Monotonic transducers have a few advantages. First, RNN-T can suffer from runaway hallucination, where a model keeps emitting non-blank symbols without advancing in time, often in an infinite loop. Secondly, monotonic transducers consume exactly one model score per time step and are therefore more compatible and unifiable with traditional FST-based hybrid ASR decoders. However, the MonoRNN-T so far has been found to have worse accuracy than RNN-T. It does not have to be that way, though: By regularizing the training - via joint LAS training or parameter initialization from RNN-T - both MonoRNN-T and CTC-T perform as well - or better - than RNN-T. This is demonstrated for LibriSpeech and for a large-scale in-house data set.

* Submitted to Interspeech 2022 
Viaarxiv icon

Federated Domain Adaptation for ASR with Full Self-Supervision

Apr 05, 2022
Junteng Jia, Jay Mahadeokar, Weiyi Zheng, Yuan Shangguan, Ozlem Kalinli, Frank Seide

Figure 1 for Federated Domain Adaptation for ASR with Full Self-Supervision
Figure 2 for Federated Domain Adaptation for ASR with Full Self-Supervision
Figure 3 for Federated Domain Adaptation for ASR with Full Self-Supervision
Figure 4 for Federated Domain Adaptation for ASR with Full Self-Supervision

Cross-device federated learning (FL) protects user privacy by collaboratively training a model on user devices, therefore eliminating the need for collecting, storing, and manually labeling user data. While important topics such as the FL training algorithm, non-IID-ness, and Differential Privacy have been well studied in the literature, this paper focuses on two challenges of practical importance for improving on-device ASR: the lack of ground-truth transcriptions and the scarcity of compute resource and network bandwidth on edge devices. First, we propose a FL system for on-device ASR domain adaptation with full self-supervision, which uses self-labeling together with data augmentation and filtering techniques. The system can improve a strong Emformer-Transducer based ASR model pretrained on out-of-domain data, using in-domain audio without any ground-truth transcriptions. Second, to reduce the training cost, we propose a self-restricted RNN Transducer (SR-RNN-T) loss, a variant of alignment-restricted RNN-T that uses Viterbi alignments from self-supervision. To further reduce the compute and network cost, we systematically explore adapting only a subset of weights in the Emformer-Transducer. Our best training recipe achieves a $12.9\%$ relative WER reduction over the strong out-of-domain baseline, which equals $70\%$ of the reduction achievable with full human supervision and centralized training.

Viaarxiv icon