SOTAVerified

Automatic Speech Recognition (ASR)

Automatic Speech Recognition (ASR) involves converting spoken language into written text. It is designed to transcribe spoken words into text in real-time, allowing people to communicate with computers, mobile devices, and other technology using their voice. The goal of Automatic Speech Recognition is to accurately transcribe speech, taking into account variations in accent, pronunciation, and speaking style, as well as background noise and other factors that can affect speech quality.

Papers

Recently AddedMost HypedMost ActiveNeeds VerificationMost Verified

Showing 351–375 of 3012 papers

TitleStatusHype
Robust ASR Error Correction with Conservative Data Filtering—0
Vibravox: A Dataset of French Speech Captured with Body-conduction Audio SensorsCode1
Beyond Binary: Multiclass Paraphasia Detection with Generative Pretrained Transformers and End-to-End Models—0
Leave No Knowledge Behind During Knowledge Distillation: Towards Practical and Effective Knowledge Distillation for Code-Switching ASR Using Realistic Data—0
Textless Dependency Parsing by Labeled Sequence PredictionCode0
Improving Neural Biasing for Contextual Speech Recognition by Early Context Injection and Text Perturbation—0
HebDB: a Weakly Supervised Dataset for Hebrew Speech Processing—0
Homogeneous Speaker Features for On-the-Fly Dysarthric and Elderly Speaker Adaptation—0
Semi-supervised Learning for Code-Switching ASR with Large Language Model Filter—0
Seed-ASR: Understanding Diverse Speech and Contexts with LLM-based Speech Recognition—0
Pretraining End-to-End Keyword Search with Automatically Discovered Acoustic UnitsCode2
LearnerVoice: A Dataset of Non-Native English Learners' Spontaneous Speech—0
Written Term Detection Improves Spoken Term DetectionCode0
Romanization Encoding For Multilingual ASR—0
Performance Analysis of Speech Encoders for Low-Resource SLU and ASR in Tunisian Dialect—0
Speculative Speech Recognition by Audio-Prefixed Low-Rank Adaptation of Language Models—0
Controlling Whisper: Universal Acoustic Adversarial Attacks to Control Speech Foundation ModelsCode1
Improving Accented Speech Recognition using Data Augmentation based on Unsupervised Text-to-Speech Synthesis—0
Improving Self-supervised Pre-training using Accent-Specific CodebooksCode1
Multi-Convformer: Extending Conformer with Multiple Convolution Kernels—0
Codec-ASR: Training Performant Automatic Speech Recognition Systems with Discrete Speech Representations—0
Pinyin Regularization in Error Correction for Chinese Speech Recognition with Large Language ModelsCode1
Error Correction by Paying Attention to Both Acoustic and Confidence References for Automatic Speech Recognition—0
Tradition or Innovation: A Comparison of Modern ASR Methods for Forced Alignment—0
Applying LLMs for Rescoring N-best ASR Hypotheses of Casual Conversations: Effects of Domain Adaptation and Context Carry-over—0
Show:102550
← PrevPage 15 of 121Next →
All datasetsLRS2RealMANSagaleeHUI speech corpusLRS3-TEDM-AILabs speech datasetThe Spoken Wikipedia CorporaVoxforge GermanVoxPopuli

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1TM-CTCTest WER10.1—Unverified
2TM-seq2seqTest WER9.7—Unverified
3CTC/attentionTest WER8.2—Unverified
4LF-MMI TDNNTest WER6.7—Unverified
5Whisper-LLaMATest WER6.6—Unverified
6End2end ConformerTest WER3.9—Unverified
7End2end ConformerTest WER3.7—Unverified
8MoCo + wav2vec (w/o extLM)Test WER2.7—Unverified
9CTC/AttentionTest WER1.5—Unverified
10WhisperTest WER1.3—Unverified
#ModelMetricClaimedVerifiedStatus
1SpatialNetCER14.5—Unverified
2CleanMel-L-maskCER14.4—Unverified
#ModelMetricClaimedVerifiedStatus
1ConformerTest WER15.32—Unverified
2Whisper-largev3-finetunedTest WER10.82—Unverified
#ModelMetricClaimedVerifiedStatus
1Conformer TransducerWER (%)1.89—Unverified
#ModelMetricClaimedVerifiedStatus
1DistillAVWER1.4—Unverified
#ModelMetricClaimedVerifiedStatus
1Conformer TransducerWER (%)4.28—Unverified
#ModelMetricClaimedVerifiedStatus
1Conformer TransducerWER (%)8.04—Unverified
#ModelMetricClaimedVerifiedStatus
1Conformer TransducerWER (%)3.36—Unverified
#ModelMetricClaimedVerifiedStatus
1Conformer Transducer (German)WER (%)8.98—Unverified