Orukeet, new ASR model based on Parakeet
Orukeet, a new ASR model based on Parakeet, has been introduced, showing improved performance. It outperforms Parakeet on 61 of 74 tested splits, including LibriSpeech test-clean (1.46% vs. 1.53% WER), test-other (2.86% vs. 3.14%), and FLEURS English (3.82% vs. 4.28%). Across all 25 FLEURS languages, Orukeet achieves a pooled WER of 9.85% compared to Parakeet's 11.01%, representing a 10.6% relative reduction.
Time & source
Times shown in UTC
Display time zone: UTC
Local time zone unavailable; showing UTC.
PublishedOffset at this time: UTC+0Sep 11, 2026, 15:50 UTC
IngestedOffset at this time: UTC+0Sep 12, 2026, 15:01 UTC
- Published
- Sep 11, 2026, 15:50
- Ingested
- Sep 12, 2026, 15:01
- Source type
- Dev community
- Tier
- Community
- Source status
- Healthy
Tier is a per-source editorial setting, not a per-item score.
Discussion trend
The percentage is based on collected discussion signal, not new comments or independent people. The curve only compares the same topic across time.
I haven't seen this mentioned yet, so I thought it deserves a post. I was trying out OpenWhispr when this model came up as the recommendation. So I don't have personal experience yet, but it's supposed to be a better version of Parakeet, especially on Macs.
Their official tidbit: "Orukeet is a 25-language speech recognizer built from NVIDIA Parakeet TDT 0.6B v3. It replaces half of the encoder's temporal depthwise filters with 12,288 fitted, frozen Gabor kernels and trains the remaining parameters on multilingual and multi-accent data.
Orukeet outperforms Parakeet on 61 of 74 tested splits, including LibriSpeech test-clean (1.46% vs. 1.53% WER), test-other (2.86% vs. 3.14%), and FLEURS English (3.82% vs. 4.28%). Across all 25 FLEURS languages, pooled WER is 9.85% vs. 11.01%, a 10.6% relative reduction. Final adaptation and checkpoint selection use LibriSpeech test-other."