Gulati and colleagues: Conformer for speech recognition

What changed

Combined convolution and self-attention in a parameter-efficient encoder for speech recognition, capturing local and global sequence patterns.

Scientific scope

Combined convolution and self-attention in a parameter-efficient encoder for speech recognition, capturing local and global sequence patterns. Scope limitation: The Conformer architecture and its speech-recognition evaluation are one result; not the original Transformer or convolutional network.

Evidence

The retained primary-source abstract, bibliographic record and stated review scope support this provisional assessment. No independent experimental replication by WRS is claimed. The Conformer architecture and its speech-recognition evaluation are one result; not the original Transformer or convolutional network.

Contributors

• The named authors listed in the linked research record · SCIENCE: Complete source byline retained in the research record; human research direction, method development and evaluation. Individual labor shares are not measured.

• Gulati and colleagues: Conformer for speech recognition — original paper
• Gulati and colleagues: Conformer for speech recognition — official company source

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction