AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

Self-supervised learning with random-projection quantizer for speech recognition

Chung-Cheng Chiu, James Qin, Yu Zhang, Jiahui Yu, Yonghui Wu

ICML 2022 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Self-supervised speech and brain decoding · 30 citations (OpenAlex)

Read the paper

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use BEST-RQ instead of wav2vec 2.0 for streaming pre-trainingConformer 0.6B · LibriSpeech testWER: 2.9 -> 2.8Improved
Use BEST-RQ instead of w2v-BERT for streaming pre-trainingConformer 0.6B · LibriSpeech test-otherWER: 8.1 -> 6.6Improved
Use BEST-RQ streaming pre-training rather than the Conformer 0.1B baselineConformer 0.6B · LibriSpeechrelative latency (ms): 0 -> -130.9Improved
Use BEST-RQ instead of w2v-BERTConformer 0.6B · MLS-10hrsaverage WER: 9.9 -> 9.6Improved
Use BEST-RQ instead of w2v-BERTConformer 0.6B · Voice Search VS-1000hrsaverage WER: 11.5 -> 10.9Improved
Replace random quantization with Projection VQ-VAE targets (ablation)BEST-RQ pretrain-finetune · LibriSpeechWER dev-other: 2.8 -> 2.8No clear effect
Replace random quantization with Transformer VQ-VAE targets (ablation)BEST-RQ pretrain-finetune · LibriSpeechWER dev-other: 2.8 -> 2.9No clear effect
Use Transformer VQ-VAE quantized tokens rather than random quantized tokens as direct ASR inputsDirect ASR · LibriSpeech 960hWER dev-other: 78.8 -> 35.8Improved
Increase pre-training data when using the random-projection quantizer (ablation)BEST-RQ · LibriSpeech fine-tuningWERImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy