Model reference · open weights
wavlm-vindata-demo-dist is an open-weight audio or speech model from phongdtd. AxForge deploys and operates it for you on dedicated EU-owned hardware — with the licence handled where one is required.
Available as managed deployment — configured and operated for you on dedicated EU hardware, quoted per deployment.
What it is
| Released by | phongdtd |
|---|---|
| Type | Audio & music |
| Task | Speech→text |
| Runs with | transformers |
| Released | 2022-03-02 |
| Popularity | 1k downloads / month |
| Licence | Unknown |
About
This model is a fine-tuned version of microsoft/wavlm-base on the PHONGDTD/VINDATAVLSP - NA dataset. It achieves the following results on the evaluation set:
More information needed
More information needed
More information needed
The following hyperparameters were used during training:
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 4.0704 | 0.01 | 100 | 3.8768 | 1.0 |
| 3.6236 | 0.01 | 200 | 3.4611 | 1.0 |
| 6.597 | 0.02 | 300 | 3.4557 | 1.0 |
| 3.4744 | 0.03 | 400 | 3.4567 | 1.0 |
| 5.3992 | 0.04 | 500 | 3.4631 | 1.0 |
| 4.5348 | 0.04 | 600 | 3.4651 | 1.0 |
| 3.2457 | 0.05 | 700 | 3.4917 | 1.0 |
| 3.9245 | 0.06 | 800 | 3.4680 | 1.0 |
| 3.2904 | 0.07 | 900 | 3.4518 | 1.0 |
| 3.4768 | 0.07 | 1000 | 3.4506 | 1.0 |
| 3.2418 | 0.08 | 1100 | 3.4474 | 1.0 |
| 3.3111 | 0.09 | 1200 | 3.4684 | 1.0 |
| 3.986 | 0.09 | 1300 | 3.4465 | 1.0 |
| 4.3206 | 0.1 | 1400 | 3.4723 | 1.0 |
| 4.682 | 0.11 | 1500 | 3.4732 | 1.0 |
| 4.858 | 0.12 | 1600 | 3.4416 | 1.0 |
| 3.2949 | 0.12 | 1700 | 3.4481 | 1.0 |
| 3.4435 | 0.13 | 1800 | 3.4570 | 1.0 |
| 5.0695 | 0.14 | 1900 | 3.4448 | 1.0 |
| 3.4962 | 0.14 | 2000 | 3.4416 | 1.0 |
| 3.4891 | 0.15 | 2100 | 3.4455 | 1.0 |
| 4.1281 | 0.16 | 2200 | 3.4447 | 1.0 |
| 3.5956 | 0.17 | 2300 | 3.4512 | 1.0 |
| 3.6312 | 0.17 | 2400 | 3.4484 | 1.0 |
| 4.5383 | 0.18 | 2500 | 3.4435 | 1.0 |
| 6.1329 | 0.19 | 2600 | 3.4530 | 1.0 |
| 3.709 | 0.2 | 2700 | 3.4466 | 1.0 |
| 3.289 | 0.2 | 2800 | 3.4463 | 1.0 |
| 4.3301 | 0.21 | 2900 | 3.4418 | 1.0 |
| 4.6656 | 0.22 | 3000 | 3.4447 | 1.0 |
| 3.4288 | 0.22 | 3100 | 3.4715 | 1.0 |
| 3.5506 | 0.23 | 3200 | 3.4437 | 1.0 |
| 3.7497 | 0.24 | 3300 | 3.4910 | 1.0 |
| 3.5198 | 0.25 | 3400 | 3.4574 | 1.0 |
| 3.4183 | 0.25 | 3500 | 3.4607 | 1.0 |
| 4.5573 | 0.26 | 3600 | 3.4421 | 1.0 |
| 3.5737 | 0.27 | 3700 | 3.4481 | 1.0 |
| 4.9008 | 0.28 | 3800 | 3.4411 | 1.0 |
| 4.8725 | 0.28 | 3900 | 3.4422 | 1.0 |
| 3.5799 | 0.29 | 4000 | 3.4659 | 1.0 |
| 3.3257 | 0.3 | 4100 | 3.4519 | 1.0 |
| 3.6887 | 0.3 | 4200 | 3.4827 | 1.0 |
| 3.3037 | 0.31 | 4300 | 3.4632 | 1.0 |
| 5.5543 | 0.32 | 4400 | 3.4480 | 1.0 |
| 3.2898 | 0.33 | 4500 | 3.4404 | 1.0 |
| 3.2794 | 0.33 | 4600 | 3.4633 | 1.0 |
| 3.7896 | 0.34 | 4700 | 3.4439 | 1.0 |
| 3.6662 | 0.35 | 4800 | 3.4587 | 1.0 |
| 3.588 | 0.35 | 4900 | 3.4520 | 1.0 |
| 4.0535 | 0.36 | 5000 | 3.4450 | 1.0 |
| 3.4335 | 0.37 | 5100 | 3.4577 | 1.0 |
| 3.6317 | 0.38 | 5200 | 3.4443 | 1.0 |
| 5.2564 | 0.38 | 5300 | 3.4505 | 1.0 |
| 3.8781 | 0.39 | 5400 | 3.4418 | 1.0 |
| 4.6269 | 0.4 | 5500 | 3.4425 | 1.0 |
| 3.6095 | 0.41 | 5600 | 3.4581 | 1.0 |
| 4.6164 | 0.41 | 5700 | 3.4404 | 1.0 |
| 3.117 | 0.42 | 5800 | 3.4596 | 1.0 |
| 4.3939 | 0.43 | 5900 | 3.4401 | 1.0 |
| 3.5856 | 0.43 | 6000 | 3.4413 | 1.0 |
| 3.5187 | 0.44 | 6100 | 3.4452 | 1.0 |
| 4.7991 | 0.45 | 6200 | 3.4481 | 1.0 |
| 3.3905 | 0.46 | 6300 | 3.4420 | 1.0 |
| 3.5086 | 0.46 | 6400 | 3.4494 | 1.0 |
| 4.8217 | 0.47 | 6500 | 3.4477 | 1.0 |
| 3.3193 | 0.48 | 6600 | 3.4382 | 1.0 |
| 5.3482 | 0.49 | 6700 | 3.4580 | 1.0 |
| 3.3947 | 0.49 | 6800 | 3.4767 | 1.0 |
| 6.3352 | 0.5 | 6900 | 3.4476 | 1.0 |
| 3.4448 | 0.51 | 7000 | 3.4557 | 1.0 |
| 3.5358 | 0.51 | 7100 | 3.4438 | 1.0 |
| 3.3499 | 0.52 | 7200 | 3.4445 | 1.0 |
| 3.6932 | 0.53 | 7300 | 3.4463 | 1.0 |
| 6.9058 | 0.54 | 7400 | 3.4482 | 1.0 |
| 4.5514 | 0.54 | 7500 | 3.4422 | 1.0 |
| 3.517 | 0.55 | 7600 | 3.4505 | 1.0 |
| 7.4479 | 0.56 | 7700 | 3.4461 | 1.0 |
| 3.3761 | 0.56 | 7800 | 3.4511 | 1.0 |
| 4.5925 | 0.57 | 7900 | 3.4389 | 1.0 |
| 5.2682 | 0.58 | 8000 | 3.4563 | 1.0 |
| 5.6748 | 0.59 | 8100 | 3.4601 | 1.0 |
| 4.4335 | 0.59 | 8200 | 3.4439 | 1.0 |
| 5.1686 | 0.6 | 8300 | 3.4444 | 1.0 |
| 3.5245 | 0.61 | 8400 | 3.4629 | 1.0 |
| 4.9426 | 0.62 | 8500 | 3.438 |
From the published model card. Full card on the HuggingFace links in the sidebar.
Using it via the API
Once AxForge deploys wavlm-vindata-demo-dist for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (wavlm-vindata-demo-dist below is illustrative; you get the exact model name on deployment.)
$ curl -sS https://api.axforge.ai/v1/audio/transcriptions \ -H "Authorization: Bearer $AXFORGE_API_KEY" \ -F model="wavlm-vindata-demo-dist" -F file=@audio.mp3
Create an account — your API key is available in the console. 3M free tokens every 30 days with every new account.