Update README.md
Browse files
README.md
CHANGED
|
@@ -150,4 +150,16 @@ language:
|
|
| 150 |
- ms
|
| 151 |
- zu
|
| 152 |
- za
|
| 153 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 150 |
- ms
|
| 151 |
- zu
|
| 152 |
- za
|
| 153 |
+
---
|
| 154 |
+
|
| 155 |
+
[XEUS - A Cross-lingual Encoder for Universal Speech]()
|
| 156 |
+
|
| 157 |
+
XEUS is a large-scale multilingual speech encoder by Carnegie Mellon University's WAVLab that covers over **4000** languages. It is pre-trained on over 1 million hours of publicly available speech datasets. It can be requires fine-tuning to be used in downstream tasks such as Speech Recognition or Translation. XEUS uses the [E-Branchformer]() architecture and is trained using [HuBERT]()-style masked prediction of discrete speech tokens. During training, the input speech is also augmented with acoustic noise and reverberation, making XEUS more robust.
|
| 158 |
+
|
| 159 |
+
XEUS tops the [ML-SUPERB]() multilingual speech recognition benchmark, outperforming [MMS](), [w2v-BERT 2.0](), and [XLS-R](). XEUS also sets a new state-of-the-art on 4 tasks in the monolingual [SUPERB]() benchmark.
|
| 160 |
+
|
| 161 |
+
|
| 162 |
+
|
| 163 |
+
|
| 164 |
+
|
| 165 |
+
|