Wav2vec2 1b Npsc Nst Bokmaal
W
Wav2vec2 1b Npsc Nst Bokmaal
Developed by NbAiLab
This model is an automatic speech recognition (ASR) model fine-tuned on the Norwegian Bokmål dialect speech dataset based on facebook/wav2vec2-xls-r-1b
Downloads 30
Release Time : 5/23/2022
Model Overview
A speech recognition model optimized for the Norwegian Bokmål dialect, fine-tuned on the wav2vec2-xls-r-1b architecture with high recognition accuracy
Model Features
High Accuracy
Achieves a low word error rate (WER) of 0.0345 on the evaluation set
Large-scale Pretraining
Fine-tuned based on the 1-billion-parameter large-scale wav2vec2-xls-r model
Norwegian Optimization
Specifically optimized for the Norwegian Bokmål dialect
Model Capabilities
Norwegian speech recognition
Speech-to-text conversion
Audio content understanding
Use Cases
Speech Transcription
Norwegian Meeting Minutes
Automatically transcribe Norwegian meeting recordings into text records
Highly accurate transcription results
Voice Assistant
Provide speech recognition capabilities for Norwegian voice assistants
Education
Language Learning Applications
Help learners practice Norwegian pronunciation and listening
🚀 wav2vec2-1b-npsc-nst-bokmaal
This model is a fine - tuned version of facebook/wav2vec2-xls-r-1b on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0636
- Wer: 0.0345
🚀 Quick Start
This model is a fine - tuned version of the base model facebook/wav2vec2-xls-r-1b. You can directly use it for relevant speech - related tasks.
📚 Documentation
Model description
This model is a fine - tuned version of facebook/wav2vec2-xls-r-1b on the None dataset.
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e - 05
- train_batch_size: 12
- eval_batch_size: 12
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 24
- optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 2000
- num_epochs: 40.0
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
2.9289 | 0.04 | 500 | 2.7971 | 1.0 |
0.7651 | 0.08 | 1000 | 0.5766 | 0.5701 |
0.518 | 0.12 | 1500 | 0.3660 | 0.3564 |
0.5481 | 0.16 | 2000 | 0.2730 | 0.2557 |
0.4286 | 0.2 | 2500 | 0.2282 | 0.2385 |
0.4543 | 0.24 | 3000 | 0.1852 | 0.1843 |
0.4064 | 0.28 | 3500 | 0.1574 | 0.1544 |
0.3308 | 0.32 | 4000 | 0.1558 | 0.1404 |
0.3005 | 0.36 | 4500 | 0.1417 | 0.1349 |
0.2898 | 0.4 | 5000 | 0.1304 | 0.1301 |
0.2739 | 0.44 | 5500 | 0.1316 | 0.1258 |
0.2429 | 0.48 | 6000 | 0.1222 | 0.1221 |
0.2405 | 0.52 | 6500 | 0.1224 | 0.1147 |
0.2681 | 0.56 | 7000 | 0.1162 | 0.1119 |
0.2382 | 0.6 | 7500 | 0.1251 | 0.1338 |
0.2193 | 0.64 | 8000 | 0.1121 | 0.1087 |
0.2235 | 0.68 | 8500 | 0.1058 | 0.1059 |
0.2124 | 0.72 | 9000 | 0.1106 | 0.1060 |
0.2076 | 0.76 | 9500 | 0.1023 | 0.1003 |
0.2082 | 0.8 | 10000 | 0.0985 | 0.1037 |
0.2272 | 0.84 | 10500 | 0.0990 | 0.1019 |
0.1903 | 0.88 | 11000 | 0.0982 | 0.0978 |
0.2044 | 0.92 | 11500 | 0.0966 | 0.0991 |
0.1976 | 0.96 | 12000 | 0.0925 | 0.0958 |
0.1966 | 1.0 | 12500 | 0.0924 | 0.0929 |
0.1959 | 1.04 | 13000 | 0.0888 | 0.0918 |
0.1727 | 1.08 | 13500 | 0.0901 | 0.0943 |
0.1853 | 1.12 | 14000 | 0.0934 | 0.0890 |
0.1791 | 1.16 | 14500 | 0.0935 | 0.0914 |
0.1872 | 1.2 | 15000 | 0.0851 | 0.0883 |
0.1821 | 1.24 | 15500 | 0.0857 | 0.0873 |
0.1756 | 1.28 | 16000 | 0.0884 | 0.0890 |
0.1666 | 1.32 | 16500 | 0.0871 | 0.0849 |
0.1708 | 1.36 | 17000 | 0.0837 | 0.0863 |
0.1653 | 1.4 | 17500 | 0.0831 | 0.0849 |
0.1734 | 1.44 | 18000 | 0.0808 | 0.0850 |
0.1643 | 1.48 | 18500 | 0.0814 | 0.0835 |
0.1635 | 1.52 | 19000 | 0.0810 | 0.0816 |
0.1611 | 1.56 | 19500 | 0.0827 | 0.0829 |
0.1633 | 1.6 | 20000 | 0.0797 | 0.0820 |
0.1626 | 1.64 | 20500 | 0.0789 | 0.0820 |
0.1618 | 1.68 | 21000 | 0.0766 | 0.0817 |
0.1634 | 1.72 | 21500 | 0.0783 | 0.0832 |
0.1761 | 1.76 | 22000 | 0.0787 | 0.0817 |
0.1518 | 1.8 | 22500 | 0.0775 | 0.0817 |
0.1492 | 1.84 | 23000 | 0.0785 | 0.0802 |
0.1652 | 1.88 | 23500 | 0.0759 | 0.0783 |
0.1545 | 1.92 | 24000 | 0.0758 | 0.0788 |
0.1548 | 1.96 | 24500 | 0.0774 | 0.0816 |
0.1641 | 2.0 | 25000 | 0.0734 | 0.0780 |
0.1506 | 2.04 | 25500 | 0.0718 | 0.0748 |
0.1474 | 2.08 | 26000 | 0.0748 | 0.0746 |
0.137 | 2.12 | 26500 | 0.0736 | 0.0744 |
0.1483 | 2.16 | 27000 | 0.0744 | 0.0763 |
0.1472 | 2.2 | 27500 | 0.0728 | 0.0736 |
0.143 | 2.24 | 28000 | 0.0708 | 0.0767 |
0.1467 | 2.28 | 28500 | 0.0721 | 0.0737 |
0.1286 | 2.32 | 29000 | 0.0701 | 0.0748 |
0.1407 | 2.36 | 29500 | 0.0695 | 0.0740 |
0.1377 | 2.4 | 30000 | 0.0674 | 0.0725 |
0.1344 | 2.44 | 30500 | 0.0696 | 0.0711 |
0.1337 | 2.48 | 31000 | 0.0686 | 0.0733 |
0.1384 | 2.52 | 31500 | 0.0686 | 0.0710 |
0.1355 | 2.56 | 32000 | 0.0667 | 0.0721 |
0.1334 | 2.6 | 32500 | 0.0665 | 0.0712 |
0.1401 | 2.64 | 33000 | 0.0694 | 0.0719 |
0.1368 | 2.68 | 33500 | 0.0689 | 0.0692 |
0.1259 | 2.72 | 34000 | 0.0669 | 0.0701 |
0.1354 | 2.76 | 34500 | 0.0672 | 0.0691 |
0.1319 | 2.8 | 35000 | 0.0707 | 0.0702 |
0.1408 | 2.84 | 35500 | 0.0650 | 0.0685 |
0.1355 | 2.88 | 36000 | 0.0671 | 0.0696 |
0.1252 | 2.92 | 36500 | 0.0655 | 0.0671 |
0.155 | 2.96 | 37000 | 0.0662 | 0.0679 |
0.1266 | 3.0 | 37500 | 0.0654 | 0.0669 |
0.1183 | 3.04 | 38000 | 0.0655 | 0.0664 |
0.1213 | 3.08 | 38500 | 0.0668 | 0.0665 |
0.1099 | 3.12 | 39000 | 0.0662 | 0.0660 |
0.1196 | 3.16 | 39500 | 0.0652 | 0.0657 |
0.1253 | 3.2 | 40000 | 0.0674 | 0.0655 |
0.1172 | 3.24 | 40500 | 0.0656 | 0.0654 |
0.1207 | 3.28 | 41000 | 0.0640 | 0.0660 |
0.1228 | 3.32 | 41500 | 0.0668 | 0.0658 |
0.1203 | 3.36 | 42000 | 0.0640 | 0.0642 |
0.1284 | 3.4 | 42500 | 0.0650 | 0.0664 |
0.1058 | 3.44 | 43000 | 0.0618 | 0.0648 |
0.1236 | 3.48 | 43500 | 0.0638 | 0.0649 |
0.1149 | 3.52 | 44000 | 0.0636 | 0.0653 |
0.1203 | 3.56 | 44500 | 0.0633 | 0.0644 |
0.114 | 3.6 | 45000 | 0.0624 | 0.0635 |
0.1201 | 3.64 | 45500 | 0.0608 | 0.0638 |
0.121 | 3.68 | 46000 | 0.0610 | 0.0631 |
0.1125 | 3.72 | 46500 | 0.0603 | 0.0634 |
0.1149 | 3.76 | 47000 | 0.0616 | 0.0644 |
0.1119 | 3.8 | 47500 | 0.0623 | 0.0638 |
0.124 | 3.84 | 48000 | 0.0610 | 0.0629 |
0.1113 | 3.88 | 48500 | 0.0612 | 0.0616 |
0.116 | 3.92 | 49000 | 0.0607 | 0.0641 |
0.1096 | 3.96 | 49500 | 0.0603 | 0.0626 |
0.1144 | 4.0 | 50000 | 0.0607 | 0.0617 |
0.1052 | 4.04 | 50500 | 0.0597 | 0.0612 |
0.1163 | 4.08 | 51000 | 0.0607 | 0.0631 |
0.1031 | 4.12 | 51500 | 0.0605 | 0.0609 |
0.1087 | 4.16 | 52000 | 0.0617 | 0.0621 |
0.1132 | 4.2 | 52500 | 0.0618 | 0.0611 |
0.0956 | 4.24 | 53000 | 0.0619 | 0.0615 |
0.1055 | 4.28 | 53500 | 0.0602 | 0.0607 |
0.1088 | 4.32 | 54000 | 0.0608 | 0.0595 |
0.1153 | 4.36 | 54500 | 0.0598 | 0.0605 |
0.0997 | 4.4 | 55000 | 0.0590 | 0.0605 |
0.1092 | 4.44 | 55500 | 0.0597 | 0.0606 |
0.1112 | 4.48 | 56000 | 0.0602 | 0.0594 |
0.0959 | 4.52 | 56500 | 0.0595 | 0.0591 |
0.1072 | 4.56 | 57000 | 0.0592 | 0.0602 |
0.1068 | 4.6 | 57500 | 0.0586 | 0.0594 |
0.111 | 4.64 | 58000 | 0.0570 | 0.0587 |
0.1 | 4.68 | 58500 | 0.0594 | 0.0596 |
0.1082 | 4.72 | 59000 | 0.0589 | 0.0590 |
0.1023 | 4.76 | 59500 | 0.0574 | 0.0590 |
0.1053 | 4.8 | 60000 | 0.0575 | 0.0588 |
0.0984 | 4.84 | 60500 | 0.0575 | 0.0583 |
0.0995 | 4.88 | 61000 | 0.0597 | 0.0591 |
0.0955 | 4.92 | 61500 | 0.0560 | 0.0590 |
0.1186 | 4.96 | 62000 | 0.0591 | 0.0577 |
0.0976 | 5.0 | 62500 | 0.0573 | 0.0585 |
0.1049 | 5.04 | 63000 | 0.0578 | 0.0572 |
0.0953 | 5.08 | 63500 | 0.0587 | 0.0574 |
0.0976 | 5.12 | 64000 | 0.0567 | 0.0589 |
0.1019 | 5.16 | 64500 | 0.0583 | 0.0582 |
0.0986 | 5.2 | 65000 | 0.0577 | 0.0571 |
0.0985 | 5.24 | 65500 | 0.0572 | 0.0577 |
0.103 | 5.28 | 66000 | 0.0581 | 0.0593 |
0.101 | 5.32 | 66500 | 0.0575 | 0.0579 |
0.1085 | 5.36 | 67000 | 0.0582 | 0.0583 |
0.1029 | 5.4 | 67500 | 0.0565 | 0.0581 |
0.0954 | 5.44 | 68000 | 0.0560 | 0.0582 |
0.0974 | 5.48 | 68500 | 0.0565 | 0.0577 |
0.0929 | 5.52 | 69000 | 0.0573 | 0.0575 |
0.099 | 5.56 | 69500 | 0.0565 | 0.0561 |
0.1009 | 5.6 | 70000 | 0.0566 | 0.0563 |
0.2268 | 5.64 | 70500 | 0.0564 | 0.0569 |
0.0974 | 5.68 | 71000 | 0.0565 | 0.0567 |
0.1101 | 5.72 | 71500 | 0.0573 | 0.0559 |
0.088 | 5.76 | 72000 | 0.0576 | 0.0551 |
0.1106 | 5.8 | 72500 | 0.0553 | 0.0559 |
0.0934 | 5.84 | 73000 | 0.0548 | 0.0561 |
0.0949 | 5.88 | 73500 | 0.0552 | 0.0560 |
0.0906 | 5.92 | 74000 | 0.0538 | 0.0570 |
0.1038 | 5.96 | 74500 | 0.0563 | 0.0563 |
0.1056 | 6.0 | 75000 | 0.0564 | 0.0556 |
0.0983 | 6.04 | 75500 | 0.0570 | 0.0560 |
0.0917 | 6.08 | 76000 | 0.0563 | 0.0560 |
0.096 | 6.12 | 76500 | 0.0558 | 0.0549 |
0.0971 | 6.16 | 77000 | 0.0569 | 0.0564 |
0.0917 | 6.2 | 77500 | 0.0569 | 0.0552 |
0.0896 | 6.24 | 78000 | 0.0568 | 0.0550 |
0.0886 | 6.28 | 78500 | 0.0550 | 0.0550 |
0.0917 | 6.32 | 79000 | 0.0554 | 0.0562 |
0.0839 | 6.36 | 79500 | 0.0551 | 0.0570 |
0.0856 | 6.4 | 80000 | 0.0533 | 0.0545 |
0.0939 | 6.44 | 80500 | 0.0564 | 0.0545 |
0.0868 | 6.48 | 81000 | 0.0556 | 0.0557 |
0.0882 | 6.52 | 81500 | 0.0547 | 0.0544 |
0.0925 | 6.56 | 82000 | 0.0577 | 0.0557 |
0.0866 | 6.6 | 82500 | 0.0534 | 0.0555 |
0.091 | 6.64 | 83000 | 0.0565 | 0.0552 |
0.1033 | 6.68 | 83500 | 0.0539 | 0.0551 |
0.0953 | 6.72 | 84000 | 0.0527 | 0.0545 |
0.0866 | 6.76 | 84500 | 0.0547 | 0.0546 |
0.0912 | 6.8 | 85000 | 0.0547 | 0.0557 |
0.0901 | 6.84 | 85500 | 0.0533 | 0.0544 |
0.0859 | 6.88 | 86000 | 0.0556 | 0.0540 |
0.2118 | 6.92 | 86500 | 0.0527 | 0.0545 |
0.0868 | 6.96 | 87000 | 0.0546 | 0.0537 |
0.097 | 7.0 | 87500 | 0.0520 | 0.0557 |
0.0835 | 7.04 | 88000 | 0.0542 | 0.0538 |
0.084 | 7.08 | 88500 | 0.0545 | 0.0543 |
0.0983 | 7.12 | 89000 | 0.0528 | 0.0557 |
0.09 | 7.16 | 89500 | 0.0542 | 0.0540 |
0.0879 | 7.2 | 90000 | 0.0559 | 0.0533 |
0.0818 | 7.24 | 90500 | 0.0546 | 0.0529 |
0.0849 | 7.28 | 91000 | 0.0535 | 0.0533 |
0.0777 | 7.32 | 91500 | 0.0548 | 0.0544 |
0.0887 | 7.36 | 92000 | 0.0545 | 0.0533 |
0.0886 | 7.4 | 92500 | 0.0545 | 0.0527 |
0.0752 | 7.44 | 93000 | 0.0552 | 0.0531 |
0.0819 | 7.48 | 93500 | 0.0525 | 0.0532 |
0.0753 | 7.52 | 94000 | 0.0522 | 0.051 |
📄 License
This model is released under the Apache - 2.0 license.
Voice Activity Detection
MIT
Voice activity detection model based on pyannote.audio 2.1, used to identify speech activity segments in audio
Speech Recognition
V
pyannote
7.7M
181
Wav2vec2 Large Xlsr 53 Portuguese
Apache-2.0
This is a fine-tuned XLSR-53 large model for Portuguese speech recognition tasks, trained on the Common Voice 6.1 dataset, supporting Portuguese speech-to-text conversion.
Speech Recognition Other
W
jonatasgrosman
4.9M
32
Whisper Large V3
Apache-2.0
Whisper is an advanced automatic speech recognition (ASR) and speech translation model proposed by OpenAI, trained on over 5 million hours of labeled data, with strong cross-dataset and cross-domain generalization capabilities.
Speech Recognition Supports Multiple Languages
W
openai
4.6M
4,321
Whisper Large V3 Turbo
MIT
Whisper is a state-of-the-art automatic speech recognition (ASR) and speech translation model developed by OpenAI, trained on over 5 million hours of labeled data, demonstrating strong generalization capabilities in zero-shot settings.
Speech Recognition
Transformers Supports Multiple Languages

W
openai
4.0M
2,317
Wav2vec2 Large Xlsr 53 Russian
Apache-2.0
A Russian speech recognition model fine-tuned from facebook/wav2vec2-large-xlsr-53, supporting 16kHz sampled audio input
Speech Recognition Other
W
jonatasgrosman
3.9M
54
Wav2vec2 Large Xlsr 53 Chinese Zh Cn
Apache-2.0
A Chinese speech recognition model fine-tuned based on facebook/wav2vec2-large-xlsr-53, supporting 16kHz sampling rate audio input.
Speech Recognition Chinese
W
jonatasgrosman
3.8M
110
Wav2vec2 Large Xlsr 53 Dutch
Apache-2.0
A Dutch speech recognition model fine-tuned based on facebook/wav2vec2-large-xlsr-53, trained on the Common Voice and CSS10 datasets, supporting 16kHz audio input.
Speech Recognition Other
W
jonatasgrosman
3.0M
12
Wav2vec2 Large Xlsr 53 Japanese
Apache-2.0
Japanese speech recognition model fine-tuned from facebook/wav2vec2-large-xlsr-53, supporting 16kHz sampling rate audio input
Speech Recognition Japanese
W
jonatasgrosman
2.9M
33
Mms 300m 1130 Forced Aligner
A text-to-audio forced alignment tool based on Hugging Face pre-trained models, supporting multiple languages with high memory efficiency
Speech Recognition
Transformers Supports Multiple Languages

M
MahmoudAshraf
2.5M
50
Wav2vec2 Large Xlsr 53 Arabic
Apache-2.0
Arabic speech recognition model fine-tuned from facebook/wav2vec2-large-xlsr-53, trained on Common Voice and Arabic speech corpus
Speech Recognition Arabic
W
jonatasgrosman
2.3M
37
Featured Recommended AI Models