whatmodel.sh / stt / STT En Conformer-Transducer Large
STT En Conformer-Transducer Large
↗ share
+ contribute
STT En Conformer-Transducer Large is an open-weights speech-to-text model from NVIDIA, released September 2022. Its published size is 0.12B parameters. Licence: cc-by-4.0.
Catalogue identity notes
Exact checkpoint nvidia/stt_en_conformer_transducer_large (NeMo). Params: card (~120M). released: HF repo creation month.
Score within the Open ASR English short-form WER (March 2026) comparison set only.
$ how it compares
The speech-to-text models ranked either side of STT En Conformer-Transducer Large, out of 176 with a score, then more from NVIDIA.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 126 |  SpeechBrain Conformer (LargeScaleASR / Loquacious) | 47±5 | local | ||
| 127 |  GigaAM Multilingual Large (CTC 600M) | 47±7 | local | ||
| 128 | STT En Conformer-Transducer Large | 46±4 | local | ||
| 129 |  ESPnet E-Branchformer (no LM) | 46±4 | local | ||
| 130 | ESPnet Multiconvformer (no LM) | 46±4 | local | ||
| more from NVIDIA | |||||
| 28 | Canary-Qwen-2.5B | 61±2 | local | ||
| 35 | Parakeet TDT 0.6B v2 | 60±1 | local | ||
| 37 | Parakeet Unified EN 0.6B | 60±4 | local | ||
see all speech-to-text models · find the ones that fit your machine
$ what it's good at
long-form transcription▁▁▄▁·▁▂█▄▁▁▁ #24 of 27 * · 53% of its score · 1 test
Trails STT En Conformer-CTC Small, STT En Conformer-Transducer Small here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Open ASR English long-form WER | 15.4% ✎ | ▁▁▅▁·▃▅█▆▁▂▂ beats 3/30 | 53% |
other tests2 not in a task · 41% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| LibriSpeech test-other WER lab | 3.7% ✎ | ▁▁▁▂▅▃▅▄█▂▂▁ beats 44/54 | 32% |
| LibriSpeech test-clean WER lab | 1.7% ✎ | ▁·▁·▁▁▁▁▃▇█▆ beats 79/100 | 9.2% |
$ where to run it
Other ways to run it
- Run it as published · needs ~1.08 GB, NVIDIA only · from huggingface.co/nvidia (official)
uvx --with 'nemo_toolkit[asr]' python -c "import nemo.collections.asr as a; print(a.models.ASRModel.from_pretrained('nvidia/stt_en_conformer_transducer_large').transcribe(['audio.wav']))"
Official weights and documentation: huggingface.co/nvidia ↗
$ how the score was worked out · Open ASR Leaderboard 53%, NVIDIA NeMo docs: ASR scores… 41%
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 Open ASR Leaderboard | 1 | 53% |
 NVIDIA NeMo docs: ASR scores tables (English Conformer / FastConformer) · lab-reported | 2 | 41% |
| prior · pull toward the middle | 6% | |
| weighted average | = 46.4 |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json