File size: 5,047 Bytes
57386d2 a282f14 57386d2 a282f14 57386d2 2c83b85 57386d2 2c83b85 57386d2 56854f8 2c83b85 57386d2 2c83b85 001b0ba 57386d2 2c83b85 57386d2 ac2503a 57386d2 453927a | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 | ---
datasets:
- espnet/yodas_owsmv4
language: multilingual
license: cc-by-4.0
tags:
- espnet
- audio
- automatic-speech-recognition
- speech-translation
- language-identification
library_name: espnet
pipeline_tag: automatic-speech-recognition
---
## Open Whisper-style Speech Model (OWSM)
OWSM aims to develop fully open speech foundation models using publicly available data and open-source toolkits, including [ESPnet](https://github.com/espnet/espnet).
Inference examples can be found on our [project page](https://www.wavlab.org/activities/2024/owsm/).
The Gradio demo is [here](https://huggingface.co/spaces/pyf98/OWSM_v3_demo).
[OWSM v4](https://arxiv.org/abs/2506.00338) is the latest version in the OWSM series, which significantly outperforms OWSM v3.1 in LID and multilingual ASR.
Additionally, OWSM v4 applies 8 times subsampling (instead of 4 times in OWSM v3.1) to the log Mel features, leading to a final resolution of 80 ms in the encoder.
When running inference, we recommend setting `maxlenratio=1.0` (default) instead of smaller values.
This repo contains a base-sized model with 102M parameters, developed by [Yifan Peng](https://pyf98.github.io/) (CMU).
It is trained on 320k hours of public speech data.
The newly curated data will be publicly released. Please stay tuned!
It supports the following speech-to-text tasks:
- Language identification
- Speech recognition
- Speech translation
- Utterance-level timestamp prediction
- Long-form recognition or translation
### OWSM series
#### Encoder-decoder OWSM
| Name | Size | Hugging Face Repo |
| :--- | ---: | :---------------- |
| OWSM v3.1 base | 101M | https://huggingface.co/espnet/owsm_v3.1_ebf_base |
| OWSM v3.1 small | 367M | https://huggingface.co/espnet/owsm_v3.1_ebf_small |
| OWSM v3.1 medium | 1.02B | https://huggingface.co/espnet/owsm_v3.1_ebf |
| OWSM v3.2 small | 367M | https://huggingface.co/espnet/owsm_v3.2 |
| OWSM v4 base | 102M | https://huggingface.co/espnet/owsm_v4_base_102M |
| OWSM v4 small | 370M | https://huggingface.co/espnet/owsm_v4_small_370M |
| OWSM v4 medium | 1.02B | https://huggingface.co/espnet/owsm_v4_medium_1B |
#### CTC-based OWSM
| Name | Size | Hugging Face Repo |
| :--- | ---: | :---------------- |
| OWSM-CTC v3.1 medium | 1.01B | https://huggingface.co/espnet/owsm_ctc_v3.1_1B |
| OWSM-CTC v3.2 medium | 1.01B | https://huggingface.co/espnet/owsm_ctc_v3.2_ft_1B |
| OWSM-CTC v4 medium | 1.01B | https://huggingface.co/espnet/owsm_ctc_v4_1B |
### Citations
#### OWSM v4
```BibTex
@inproceedings{owsm-v4,
title={{OWSM} v4: Improving Open Whisper-Style Speech Models via Data Scaling and Cleaning},
author={Yifan Peng and Shakeel Muhammad and Yui Sudo and William Chen and Jinchuan Tian and Chyi-Jiunn Lin and Shinji Watanabe},
booktitle={Proceedings of the Annual Conference of the International Speech Communication Association (INTERSPEECH) (accepted)},
year={2025},
}
```
#### OWSM-CTC
```BibTex
@inproceedings{owsm-ctc,
title = "{OWSM}-{CTC}: An Open Encoder-Only Speech Foundation Model for Speech Recognition, Translation, and Language Identification",
author = "Peng, Yifan and
Sudo, Yui and
Shakeel, Muhammad and
Watanabe, Shinji",
booktitle = "Proceedings of the Annual Meeting of the Association for Computational Linguistics (ACL)",
year = "2024",
month= {8},
url = "https://aclanthology.org/2024.acl-long.549",
}
```
#### OWSM v3.1 and v3.2
```BibTex
@inproceedings{owsm-v32,
title={On the Effects of Heterogeneous Data Sources on Speech-to-Text Foundation Models},
author={Jinchuan Tian and Yifan Peng and William Chen and Kwanghee Choi and Karen Livescu and Shinji Watanabe},
booktitle={Proceedings of the Annual Conference of the International Speech Communication Association (INTERSPEECH)},
year={2024},
month={9},
pdf="https://arxiv.org/pdf/2406.09282"
}
@inproceedings{owsm-v31,
title={{OWSM v3.1: Better and Faster Open Whisper-Style Speech Models based on E-Branchformer}},
author={Yifan Peng and Jinchuan Tian and William Chen and Siddhant Arora and Brian Yan and Yui Sudo and Muhammad Shakeel and Kwanghee Choi and Jiatong Shi and Xuankai Chang and Jee-weon Jung and Shinji Watanabe},
booktitle={Proceedings of the Annual Conference of the International Speech Communication Association (INTERSPEECH)},
year={2024},
month={9},
pdf="https://arxiv.org/pdf/2401.16658",
}
```
#### Initial OWSM (v1, v2, v3)
```BibTex
@inproceedings{owsm,
title={Reproducing Whisper-Style Training Using An Open-Source Toolkit And Publicly Available Data},
author={Yifan Peng and Jinchuan Tian and Brian Yan and Dan Berrebbi and Xuankai Chang and Xinjian Li and Jiatong Shi and Siddhant Arora and William Chen and Roshan Sharma and Wangyou Zhang and Yui Sudo and Muhammad Shakeel and Jee-weon Jung and Soumi Maiti and Shinji Watanabe},
booktitle={Proceedings of the IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)},
year={2023},
month={12},
pdf="https://arxiv.org/pdf/2309.13876",
}
``` |