Downloads · 30 days
104
2% of all-time downloads
wav2vec2-pretrained-clsril-23-10k
Harveenchadha/wav2vec2-pretrained-clsril-23-10k
wav2vec2-pretrained-clsril-23-10k is a feature extraction model from Harveenchadha. Use it when you need embeddings to search or compare text. It is set up for transformers.
We present a CLSRIL-23 (Cross Lingual Speech Representations on Indic Languages), a self supervised learning based audio pre-trained model which learns cross lingual speech representations from raw audio across 23 Ind…
All-time downloads
6.3K
Public
Repo size
755 MB
Likes
6
Public
Repo makeup
Click a slice to open those files.
.bin378 MB · 100%
At a glance
- Task
- Feature Extraction
- Library
- transformers
- Model type
- wav2vec2
- Access
- Public
- Created
- Mar 2, 2022
- Updated
- Aug 6, 2021
- SHA
- 026bd5f2
Model card
From the Hugging Face model README
Overview
We present a CLSRIL-23 (Cross Lingual Speech Representations on Indic Languages), a self supervised learning based audio pre-trained model which learns cross lingual speech representations from raw audio across 23 Indic languages. It is built on top of wav2vec 2.0 which is solved by training a contrastive task over masked latent speech representations and jointly learns the quantization of latents shared across all languages.
Original Repo contains models in fairseq format.
Languages in the pretraining dataset
| Language | Data (In Hrs) |
|---|---|
| Assamese | 254.9 |
| Bengali | 331.3 |
| Bodo | 26.9 |
| Dogri | 17.1 |
| English | 819.7 |
| Gujarati | 336.7 |
| Hindi | 4563.7 |
| Kannada | 451.8 |
| Kashmiri | 67.8 |
| Konkani | 36.8 |
| Maithili | 113.8 |
| Malayalam | 297.7 |
| Manipuri | 171.9 |
| Marathi | 458.2 |
| Nepali | 31.6 |
| Odia | 131.4 |
| Punjabi | 486.05 |
| Sanskrit | 58.8 |
| Santali | 6.56 |
| Sindhi | 16 |
| Tamil | 542.6 |
| Telugu | 302.8 |
| Urdu | 259.68 |
Repo for training:
Experimentation platform built on top of fairseq.
- Task
- Feature Extraction
- Library
- transformers
- Type
- wav2vec2
- Created
- Mar 2, 2022
- Updated
- Aug 6, 2021