ML Research Wiki / Benchmarks / Speech Recognition / SPGISpeech

SPGISpeech

Speech Recognition Benchmark

Performance Over Time

📊 Showing 2 results | 📏 Metric: Word Error Rate (WER)

Click "Edit" next to any result to modify it, or add a new result at the bottom. All changes will be reviewed before going live.

Yellow rows = Pending edits Green rows = Pending new results

Model	Paper	Word Error Rate (WER)	Date	Actions
Conformer	SPGISpeech: 5,000 hours of transcribed financial …	5.70	2021-04-05
parakeet-rnnt-1.1b	Fast Conformer with Linearly Scalable Attention f…	3.11	2023-05-08

Rank	Model	Paper	Word Error Rate (WER)	Date	Code
1	Conformer	SPGISpeech: 5,000 hours of transcribed financial audio for fully formatted end-to-end speech recognition	5.70	2021-04-05	📦 espnet/espnet
2	parakeet-rnnt-1.1b 📚	Fast Conformer with Linearly Scalable Attention for Efficient Speech Recognition	3.11	2023-05-08	-

2021

Conformer

espnet/espnet

2023

parakeet-rnnt-1.1b