Automatic Speech Recognition
Transformers
Safetensors
Finnish
wav2vec2
finnish
Eval Results (legacy)
Instructions to use GetmanY1/wav2vec2-base-fi-150k-finetuned with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use GetmanY1/wav2vec2-base-fi-150k-finetuned with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="GetmanY1/wav2vec2-base-fi-150k-finetuned")# Load model directly from transformers import AutoProcessor, AutoModelForCTC processor = AutoProcessor.from_pretrained("GetmanY1/wav2vec2-base-fi-150k-finetuned") model = AutoModelForCTC.from_pretrained("GetmanY1/wav2vec2-base-fi-150k-finetuned", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Update README.md
Browse files
README.md
CHANGED
|
@@ -113,7 +113,7 @@ The Finnish Wav2Vec2 Base has the same architecture and uses the same training o
|
|
| 113 |
|
| 114 |
[GetmanY1/wav2vec2-base-fi-150k](https://huggingface.co/GetmanY1/wav2vec2-base-fi-150k) is a large-scale, 95-million parameter monolingual model pre-trained on 158k hours of unlabeled Finnish speech, including [KAVI radio and television archive materials](https://kavi.fi/en/radio-ja-televisioarkistointia-vuodesta-2008/), Lahjoita puhetta (Donate Speech), Finnish Parliament, Finnish VoxPopuli.
|
| 115 |
|
| 116 |
-
You can read more about the pre-trained model from [this paper](
|
| 117 |
|
| 118 |
## Intended uses
|
| 119 |
|
|
@@ -146,6 +146,22 @@ predicted_ids = torch.argmax(logits, dim=-1)
|
|
| 146 |
transcription = processor.batch_decode(predicted_ids)
|
| 147 |
```
|
| 148 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 149 |
## Team Members
|
| 150 |
|
| 151 |
- Yaroslav Getman, [Hugging Face profile](https://huggingface.co/GetmanY1), [LinkedIn profile](https://www.linkedin.com/in/yaroslav-getman/)
|
|
|
|
| 113 |
|
| 114 |
[GetmanY1/wav2vec2-base-fi-150k](https://huggingface.co/GetmanY1/wav2vec2-base-fi-150k) is a large-scale, 95-million parameter monolingual model pre-trained on 158k hours of unlabeled Finnish speech, including [KAVI radio and television archive materials](https://kavi.fi/en/radio-ja-televisioarkistointia-vuodesta-2008/), Lahjoita puhetta (Donate Speech), Finnish Parliament, Finnish VoxPopuli.
|
| 115 |
|
| 116 |
+
You can read more about the pre-trained model from [this paper](https://www.isca-archive.org/interspeech_2025/getman25_interspeech.html). The training scripts are available on [GitHub](https://github.com/aalto-speech/large-scale-monolingual-speech-foundation-models).
|
| 117 |
|
| 118 |
## Intended uses
|
| 119 |
|
|
|
|
| 146 |
transcription = processor.batch_decode(predicted_ids)
|
| 147 |
```
|
| 148 |
|
| 149 |
+
## Citation
|
| 150 |
+
|
| 151 |
+
If you use our models or scripts, please cite our article as:
|
| 152 |
+
|
| 153 |
+
```bibtex
|
| 154 |
+
@inproceedings{getman25_interspeech,
|
| 155 |
+
title = {{Is your model big enough? Training and interpreting large-scale monolingual speech foundation models}},
|
| 156 |
+
author = {{Yaroslav Getman and Tam谩s Gr贸sz and Tommi Lehtonen and Mikko Kurimo}},
|
| 157 |
+
year = {{2025}},
|
| 158 |
+
booktitle = {{Interspeech 2025}},
|
| 159 |
+
pages = {{231--235}},
|
| 160 |
+
doi = {{10.21437/Interspeech.2025-46}},
|
| 161 |
+
issn = {{2958-1796}},
|
| 162 |
+
}
|
| 163 |
+
```
|
| 164 |
+
|
| 165 |
## Team Members
|
| 166 |
|
| 167 |
- Yaroslav Getman, [Hugging Face profile](https://huggingface.co/GetmanY1), [LinkedIn profile](https://www.linkedin.com/in/yaroslav-getman/)
|