patrickvonplaten commited on
Commit
50ff86c
1 Parent(s): 4f889aa

add readme

Browse files
Files changed (1) hide show
  1. README.md +22 -2
README.md CHANGED
@@ -1,6 +1,26 @@
1
  ---
 
2
  tags:
3
- - correct
 
 
 
 
 
 
4
  ---
5
 
6
- Test
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ language: pl
3
  tags:
4
+ - audio
5
+ - automatic-speech-recognition
6
+ - voxpopuli
7
+ datasets:
8
+ - voxpopuli
9
+ license: cc-by-nc-4.0
10
+ inference: false
11
  ---
12
 
13
+ # Wav2Vec2-base-VoxPopuli
14
+
15
+ [Facebook's Wav2Vec2](https://ai.facebook.com/blog/wav2vec-20-learning-the-structure-of-speech-from-raw-audio/) base model pretrained only in **pl** on **21.2k** unlabeled datat of the [VoxPopuli corpus](https://arxiv.org/abs/2101.00390).
16
+
17
+ The model is pretrained on 16kHz sampled speech audio. When using the model make sure that your speech input is also sampled at 16Khz.
18
+
19
+ **Note**: This model does not have a tokenizer as it was pretrained on audio alone. In order to use this model for **speech recognition**, a tokenizer should be created and the model should be fine-tuned on labeled text data in **pl**. Check out [this blog](https://huggingface.co/blog/fine-tune-xlsr-wav2vec2) for a more in-detail explanation of how to fine-tune the model.
20
+
21
+ **Paper**: *[VoxPopuli: A Large-Scale Multilingual Speech Corpus for Representation
22
+ Learning, Semi-Supervised Learning and Interpretation](https://arxiv.org/abs/2101.00390)*
23
+
24
+ **Authors**: *Changhan Wang, Morgane Riviere, Ann Lee, Anne Wu, Chaitanya Talnikar, Daniel Haziza, Mary Williamson, Juan Pino, Emmanuel Dupoux* from *Facebook AI*.
25
+
26
+ See the official website for more information, [here](https://github.com/facebookresearch/voxpopuli/).