File size: 955 Bytes
dab707d
 
68348fd
 
 
 
 
 
eb974bb
 
 
dab707d
68348fd
 
 
 
 
 
 
 
 
 
6e314d5
68348fd
6e314d5
 
68348fd
6e314d5
 
 
68348fd
 
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
---
license: mit
datasets:
- mozilla-foundation/common_voice_13_0
language:
- ca
- ta
- th
tags:
- automatic-speech-recognition
inference: false
---

## About

Multilingual Distilwhisper allows for better ASR performance in target languages by adding lightweight CLSR modules on top of whisper-small. 
These modules are trained on a mix of cross-entropy (ASR) and knowledge distillation losses, where whisper-large-v2 is used as teacher. 

## Inference

Loader will be made available soon at https://github.com/naver

## Citation
```
@inproceedings{ferraz2024distilwhisper,
  title={Multilingual DistilWhisper: Efficient Distillation of Multi-task Speech Models via Language-Specific Experts},
  author={Ferraz, Thomas Palmeira and Boito, Marcely Zanon and Brun, Caroline and Nikoulina, Vassilina},
  booktitle={ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  year={2024},
  organization={IEEE}
}
```